{"markup":"\u003C?xml version=\u00221.0\u0022 encoding=\u0022UTF-8\u0022 ?\u003E\n    \u003Chtml version=\u0022HTML+RDFa+MathML 1.1\u0022\n    xmlns:content=\u0022http:\/\/purl.org\/rss\/1.0\/modules\/content\/\u0022\n    xmlns:dc=\u0022http:\/\/purl.org\/dc\/terms\/\u0022\n    xmlns:foaf=\u0022http:\/\/xmlns.com\/foaf\/0.1\/\u0022\n    xmlns:og=\u0022http:\/\/ogp.me\/ns#\u0022\n    xmlns:rdfs=\u0022http:\/\/www.w3.org\/2000\/01\/rdf-schema#\u0022\n    xmlns:sioc=\u0022http:\/\/rdfs.org\/sioc\/ns#\u0022\n    xmlns:sioct=\u0022http:\/\/rdfs.org\/sioc\/types#\u0022\n    xmlns:skos=\u0022http:\/\/www.w3.org\/2004\/02\/skos\/core#\u0022\n    xmlns:xsd=\u0022http:\/\/www.w3.org\/2001\/XMLSchema#\u0022\n    xmlns:mml=\u0022http:\/\/www.w3.org\/1998\/Math\/MathML\u0022\u003E\n  \u003Chead\u003E\u003Cscript type=\u0022text\/javascript\u0022 src=\u0022\/\/cdn.jsdelivr.net\/qtip2\/2.2.1\/jquery.qtip.min.js\u0022\u003E\u003C\/script\u003E\n\u003Cscript type=\u0022text\/javascript\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/js\/js_YjAJQgxDlFX6S-O02jj9jCrVbrwlY3CGgCg1FzPlvBs.js\u0022\u003E\u003C\/script\u003E\n\u003Cscript type=\u0022text\/javascript\u0022\u003E\n\u003C!--\/\/--\u003E\u003C![CDATA[\/\/\u003E\u003C!--\nif(typeof window.MathJax === \u0022undefined\u0022) window.MathJax = { menuSettings: { zoom: \u0022Click\u0022 } };\n\/\/--\u003E\u003C!]]\u003E\n\u003C\/script\u003E\n\u003Cscript type=\u0022text\/javascript\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/js\/js_waP91NpgGpectm_6Y2XDEauLJ8WCSCBKmmA87unpp2E.js\u0022\u003E\u003C\/script\u003E\n\u003Cscript type=\u0022text\/javascript\u0022 src=\u0022https:\/\/www.googletagmanager.com\/gtag\/js?id=G-0K57TCX5BY\u0022\u003E\u003C\/script\u003E\n\u003Cscript type=\u0022text\/javascript\u0022\u003E\n\u003C!--\/\/--\u003E\u003C![CDATA[\/\/\u003E\u003C!--\nwindow.dataLayer = window.dataLayer || [];function gtag(){dataLayer.push(arguments)};gtag(\u0022js\u0022, new Date());gtag(\u0022set\u0022, \u0022developer_id.dMDhkMT\u0022, true);gtag(\u0022config\u0022, \u0022G-0K57TCX5BY\u0022, {\u0022groups\u0022:\u0022default\u0022,\u0022anonymize_ip\u0022:true});\n\/\/--\u003E\u003C!]]\u003E\n\u003C\/script\u003E\n\u003Cscript type=\u0022text\/javascript\u0022\u003E\n\u003C!--\/\/--\u003E\u003C![CDATA[\/\/\u003E\u003C!--\njQuery.extend(Drupal.settings, {\u0022basePath\u0022:\u0022\\\/\u0022,\u0022pathPrefix\u0022:\u0022\u0022,\u0022highwire\u0022:{\u0022ac\u0022:{\u0022medrxiv;2021.03.23.21254155v4\u0022:{\u0022access\u0022:{\u0022full\u0022:true},\u0022pisa_id\u0022:\u0022medrxiv;2021.03.23.21254155v4\u0022,\u0022apath\u0022:\u0022\u0022,\u0022jcode\u0022:\u0022medrxiv\u0022}},\u0022processed\u0022:[\u0022highwire_math\u0022],\u0022markup\u0022:[{\u0022requested\u0022:\u0022full-text\u0022,\u0022variant\u0022:\u0022full-text\u0022,\u0022view\u0022:\u0022full\u0022,\u0022pisa\u0022:\u0022medrxiv;2021.03.23.21254155v4\u0022}]},\u0022instances\u0022:\u0022{\\u0022highwire_abstract_tooltip\\u0022:{\\u0022content\\u0022:{\\u0022text\\u0022:\\u0022\\u0022},\\u0022style\\u0022:{\\u0022tip\\u0022:{\\u0022width\\u0022:20,\\u0022height\\u0022:20,\\u0022border\\u0022:1,\\u0022offset\\u0022:0,\\u0022corner\\u0022:true},\\u0022classes\\u0022:\\u0022qtip-custom hw-tooltip hw-abstract-tooltip qtip-shadow qtip-rounded\\u0022,\\u0022classes_custom\\u0022:\\u0022hw-tooltip hw-abstract-tooltip\\u0022},\\u0022position\\u0022:{\\u0022at\\u0022:\\u0022right center\\u0022,\\u0022my\\u0022:\\u0022left center\\u0022,\\u0022viewport\\u0022:true,\\u0022adjust\\u0022:{\\u0022method\\u0022:\\u0022shift\\u0022}},\\u0022show\\u0022:{\\u0022event\\u0022:\\u0022mouseenter click \\u0022,\\u0022solo\\u0022:true},\\u0022hide\\u0022:{\\u0022event\\u0022:\\u0022mouseleave \\u0022,\\u0022fixed\\u0022:1,\\u0022delay\\u0022:\\u0022100\\u0022}},\\u0022highwire_author_tooltip\\u0022:{\\u0022content\\u0022:{\\u0022text\\u0022:\\u0022\\u0022},\\u0022style\\u0022:{\\u0022tip\\u0022:{\\u0022width\\u0022:15,\\u0022height\\u0022:15,\\u0022border\\u0022:1,\\u0022offset\\u0022:0,\\u0022corner\\u0022:true},\\u0022classes\\u0022:\\u0022qtip-custom hw-tooltip hw-author-tooltip qtip-shadow qtip-rounded\\u0022,\\u0022classes_custom\\u0022:\\u0022hw-tooltip hw-author-tooltip\\u0022},\\u0022position\\u0022:{\\u0022at\\u0022:\\u0022top center\\u0022,\\u0022my\\u0022:\\u0022bottom center\\u0022,\\u0022viewport\\u0022:true,\\u0022adjust\\u0022:{\\u0022method\\u0022:\\u0022\\u0022}},\\u0022show\\u0022:{\\u0022event\\u0022:\\u0022mouseenter \\u0022,\\u0022solo\\u0022:true},\\u0022hide\\u0022:{\\u0022event\\u0022:\\u0022mouseleave \\u0022,\\u0022fixed\\u0022:1,\\u0022delay\\u0022:\\u0022100\\u0022}},\\u0022highwire_reflinks_tooltip\\u0022:{\\u0022content\\u0022:{\\u0022text\\u0022:\\u0022\\u0022},\\u0022style\\u0022:{\\u0022tip\\u0022:{\\u0022width\\u0022:15,\\u0022height\\u0022:15,\\u0022border\\u0022:1,\\u0022mimic\\u0022:\\u0022top center\\u0022,\\u0022offset\\u0022:0,\\u0022corner\\u0022:true},\\u0022classes\\u0022:\\u0022qtip-custom hw-tooltip hw-ref-link-tooltip qtip-shadow qtip-rounded\\u0022,\\u0022classes_custom\\u0022:\\u0022hw-tooltip hw-ref-link-tooltip\\u0022},\\u0022position\\u0022:{\\u0022at\\u0022:\\u0022bottom left\\u0022,\\u0022my\\u0022:\\u0022top left\\u0022,\\u0022viewport\\u0022:true,\\u0022adjust\\u0022:{\\u0022method\\u0022:\\u0022flip\\u0022}},\\u0022show\\u0022:{\\u0022event\\u0022:\\u0022mouseenter \\u0022,\\u0022solo\\u0022:true},\\u0022hide\\u0022:{\\u0022event\\u0022:\\u0022mouseleave \\u0022,\\u0022fixed\\u0022:1,\\u0022delay\\u0022:\\u0022100\\u0022}}}\u0022,\u0022qtipDebug\u0022:\u0022{\\u0022leaveElement\\u0022:0}\u0022,\u0022googleanalytics\u0022:{\u0022account\u0022:[\u0022G-0K57TCX5BY\u0022],\u0022trackOutbound\u0022:1,\u0022trackMailto\u0022:1,\u0022trackDownload\u0022:1,\u0022trackDownloadExtensions\u0022:\u00227z|aac|arc|arj|asf|asx|avi|bin|csv|doc(x|m)?|dot(x|m)?|exe|flv|gif|gz|gzip|hqx|jar|jpe?g|js|mp(2|3|4|e?g)|mov(ie)?|msi|msp|pdf|phps|png|ppt(x|m)?|pot(x|m)?|pps(x|m)?|ppam|sld(x|m)?|thmx|qtm?|ra(m|r)?|sea|sit|tar|tgz|torrent|txt|wav|wma|wmv|wpd|xls(x|m|b)?|xlt(x|m)|xlam|xml|z|zip\u0022,\u0022trackColorbox\u0022:1},\u0022ajaxPageState\u0022:{\u0022js\u0022:{\u0022\\\/\\\/cdn.jsdelivr.net\\\/qtip2\\\/2.2.1\\\/jquery.qtip.min.js\u0022:1,\u0022sites\\\/all\\\/modules\\\/highwire\\\/highwire\\\/plugins\\\/highwire_markup_process\\\/js\\\/highwire_article_reference_popup.js\u0022:1,\u0022sites\\\/all\\\/modules\\\/highwire\\\/highwire\\\/plugins\\\/highwire_markup_process\\\/js\\\/highwire_at_symbol.js\u0022:1,\u00220\u0022:1,\u0022sites\\\/all\\\/modules\\\/contrib\\\/google_analytics\\\/googleanalytics.js\u0022:1,\u0022https:\\\/\\\/www.googletagmanager.com\\\/gtag\\\/js?id=G-0K57TCX5BY\u0022:1,\u00221\u0022:1}}});\n\/\/--\u003E\u003C!]]\u003E\n\u003C\/script\u003E\n\u003Clink type=\u0022text\/css\u0022 rel=\u0022stylesheet\u0022 href=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/advagg_css\/css__dn-cpI1YtkU_iLHgA5WhlkxgYWyat_IxjF_B-WSYrpE__a9hIbt0eaZ7d5nhwnm2weG8R_2eXK4EvoOx9dOxouHE__QrrGUc7CpljPR5Aph-ukPbcwtK4AWrHGwCEXJ_k1V_c.css\u0022 media=\u0022all\u0022 \/\u003E\n\u003Clink type=\u0022text\/css\u0022 rel=\u0022stylesheet\u0022 href=\u0022\/\/cdn.jsdelivr.net\/qtip2\/2.2.1\/jquery.qtip.min.css\u0022 media=\u0022all\u0022 \/\u003E\n\u003Clink type=\u0022text\/css\u0022 rel=\u0022stylesheet\u0022 href=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/advagg_css\/css__HGACIFBlu2o05y3afvqlt5wrE_5Dn6MXsexfuEpeIwg__t4SOPxucAPoV3Os7g8dXqyMB1HRXQridRJ82X7nE33E__QrrGUc7CpljPR5Aph-ukPbcwtK4AWrHGwCEXJ_k1V_c.css\u0022 media=\u0022all\u0022 \/\u003E\n\u003Clink rel=\u0027stylesheet\u0027 type=\u0027text\/css\u0027 href=\u0027\/sites\/all\/modules\/contrib\/panels\/plugins\/layouts\/onecol\/onecol.css\u0027 \/\u003E\u003C\/head\u003E\u003Cbody\u003E\u003Cdiv class=\u0022panels-ajax-tab-panel panels-ajax-tab-panel-article-tab-full-text\u0022\u003E\u003Cdiv class=\u0022panel-display panel-1col clearfix\u0022 \u003E\n  \u003Cdiv class=\u0022panel-panel panel-col\u0022\u003E\n    \u003Cdiv\u003E\u003Cdiv class=\u0022panel-pane pane-highwire-markup\u0022 \u003E\n  \n      \n  \n  \u003Cdiv class=\u0022pane-content\u0022\u003E\n    \u003Cdiv class=\u0022highwire-markup\u0022\u003E\u003Cdiv xmlns=\u0022http:\/\/www.w3.org\/1999\/xhtml\u0022 data-highwire-cite-ref-tooltip-instance=\u0022highwire_reflinks_tooltip\u0022 class=\u0022content-block-markup\u0022 xmlns:xhtml=\u0022http:\/\/www.w3.org\/1999\/xhtml\u0022\u003E\u003Cdiv class=\u0022article fulltext-view \u0022\u003E\u003Cspan class=\u0022highwire-journal-article-marker-start\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022section abstract\u0022 id=\u0022abstract-1\u0022\u003E\u003Ch2\u003EAbstract\u003C\/h2\u003E\u003Cdiv id=\u0022sec-1\u0022 class=\u0022subsection\u0022\u003E\u003Cp id=\u0022p-7\u0022\u003E\u003Cstrong\u003EProblem definition\u003C\/strong\u003E To respond to pandemics such as COVID-19, policy makers have relied on interventions that target specific population groups or activities. Since targeting is potentially contentious, rigorously quantifying its benefits is critical for designing effective and equitable pandemic control policies.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-2\u0022 class=\u0022subsection\u0022\u003E\u003Cp id=\u0022p-8\u0022\u003E\u003Cstrong\u003EMethodology\/results\u003C\/strong\u003E We propose a flexible modeling framework and a set of associated algorithms that compute \u003Cem\u003Eoptimally targeted\u003C\/em\u003E, time-dependent interventions that coordinate across \u003Cem\u003Etwo\u003C\/em\u003E dimensions of heterogeneity: age of different groups \u003Cem\u003Eand\u003C\/em\u003E the specific activities that individuals engage in during the normal course of a day. We showcase a complete implementation focused on the \u00cele-de-France region of France, based on commonly available public data. We find that targeted policies generate substantial complementarities that lead to Pareto improvements, reducing the number of deaths and the economic losses, as well as the time in confinement for each age group. Optimized dual-targeted policies are \u003Cem\u003Einterpretable\u003C\/em\u003E: by fitting decision trees to our raw policy\u2019s decisions across many problem instances, we find that a feature corresponding to the ratio of marginal economic value prorated by social contacts is highly salient in explaining the confinements that any group - activity pair experiences. We also quantify the impact of fairness requirements that explicitly limit the differential treatment of distinct groups, and find that satisfactory trade-offs are achievable through limited targeting.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-3\u0022 class=\u0022subsection\u0022\u003E\u003Cp id=\u0022p-9\u0022\u003E\u003Cstrong\u003EImplications\u003C\/strong\u003E Given that some amount of targeting of activities \u003Cem\u003Eand\u003C\/em\u003E age groups is already in place in real-world pandemic responses, our framework highlights the significant benefits in explicitly and transparently modelling targeting and identifying the interventions that rigorously optimize overall societal welfare.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-4\u0022 class=\u0022subsection\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-5\u0022\u003E\u003Ch2 class=\u0022\u0022\u003E1. Introduction\u003C\/h2\u003E\u003Cp id=\u0022p-22\u0022\u003EThe COVID-19 pandemic has forced policy makers worldwide to rely on a range of large-scale population confinement measures in an effort to contain disease spread. In determining these measures, a key recognition has been that substantial differences exist in the health and economic impact produced by different individuals engaged in distinct activities. Targeting confinements to account for such heterogeneity could be an important lever to mitigate a pandemic\u2019s impact, but could also lead to potentially contentious and discriminatory measures. This work is aimed at developing a rigorous framework to quantify the benefits and downsides of such targeted interventions in pandemic management, and applying it to the COVID-19 pandemic as a real-world case study.\u003C\/p\u003E\u003Cp id=\u0022p-23\u0022\u003ETargeting has been implemented in several different ways during the COVID-19 pandemic. One real-world contentious example has been to differentiate confinements based on \u003Cem\u003Eage groups\u003C\/em\u003E, e.g., sheltering older individuals who might face higher health risks if infected, or restricting younger groups who might create higher infection risks. Such measures have been implemented in several settings \u2013 e.g., with stricter confinements applied to older groups in Finland (\u003Ca id=\u0022xref-ref-46-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-46\u0022\u003ETiirinki et al. 2020\u003C\/a\u003E), Ireland (\u003Ca id=\u0022xref-ref-28-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-28\u0022\u003EHarrison 2020\u003C\/a\u003E), Israel (\u003Ca id=\u0022xref-ref-34-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-34\u0022\u003EMagid 2020\u003C\/a\u003E) and Moscow (\u003Ca id=\u0022xref-ref-22-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-22\u0022\u003EFoy 2020\u003C\/a\u003E), or curfews applied to children and youth in Bosnia and Herzegovina (\u003Ca id=\u0022xref-ref-43-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-43\u0022\u003EReuters Staff 2020\u003C\/a\u003E) and Turkey (\u003Ca id=\u0022xref-ref-30-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-30\u0022\u003EKanbur and Ankg\u00fcl 2020\u003C\/a\u003E) \u2013 but some of the measures were deemed ageist and unconstitutional and were eventually overturned (\u003Ca id=\u0022xref-ref-34-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-34\u0022\u003EMagid 2020\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-43-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-43\u0022\u003EReuters Staff 2020\u003C\/a\u003E).\u003C\/p\u003E\u003Cp id=\u0022p-24\u0022\u003EA different example of targeting extensively employed in practice has been to tailor confinements to specific \u003Cem\u003Eactivities\u003C\/em\u003E conducted during a typical day. This has been driven by the recognition that different activities (or more specifically, population interactions in locations of certain activities) such as work, schooling, transport or leisure can result in significantly different patterns of social contacts and new infections. This heterogeneity has been recognized in numerous implementations that differentially confine activities through restrictions of varying degrees on schools, workplaces, recreation venues, retail spaces, etc. Additionally, some practical implementations even differentiated based on \u003Cem\u003Eboth\u003C\/em\u003E age groups and activities, e.g., by setting aside dedicated hours when only the senior population was allowed to shop at supermarkets (\u003Ca id=\u0022xref-ref-2-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-2\u0022\u003EAguilera 2020\u003C\/a\u003E), or by restricting only higher age groups from in-person work activities (\u003Ca id=\u0022xref-ref-34-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-34\u0022\u003EMagid 2020\u003C\/a\u003E).\u003C\/p\u003E\u003Cp id=\u0022p-25\u0022\u003EAs these examples suggest, targeted interventions have merits but also pose potentially significant downsides. On the one hand, targeting can generate improvements in both health and economic outcomes, giving policy makers an improved lever when navigating difficult trade-offs. Additionally, explicitly considering \u003Cem\u003Emultiple\u003C\/em\u003E dimensions of targeting simultaneously \u2013 such as activities \u003Cem\u003Eand\u003C\/em\u003E age groups \u2013 could overturn some of the prevailing insight that specific age groups should uniformly face stricter confinements. However, such granular policies are more difficult to implement, and could lead to discriminatory and potentially unfair measures.\u003C\/p\u003E\u003Cp id=\u0022p-26\u0022\u003EGiven that some amount of targeting is already in place in existing real-world policy implementations, it is critical to transparently model and quantify its benefits and downsides. This gives rise to several natural research questions: How large are the health and economic benefits of interventions that can engage in progressively finer targeting? Does finer targeting lead to significant synergies, and is there an interpretable mechanism through which this happens? What is the relationship between the effectiveness and the level of targeting allowed across distinct groups?\u003C\/p\u003E\u003Cp id=\u0022p-27\u0022\u003EWe develop a rigorous framework and a set of associated algorithms that allow addressing these questions in concrete, real-world settings. Although the framework is flexible and allows embedding different interventions, we focus on \u003Cem\u003Edual-targeted confinements\u003C\/em\u003E, that is, confinements that can target \u003Cem\u003Eage groups\u003C\/em\u003E and \u003Cem\u003Eactivities\u003C\/em\u003E, as these have been the most prevalent forms of targeting seen in practice.\u003C\/p\u003E\u003Cdiv id=\u0022sec-6\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E1.1. Contributions\u003C\/h3\u003E\u003Cp id=\u0022p-28\u0022\u003EOur framework overcomes two practical challenges. First, a rigorous quantification of the value of targeting must be grounded in formal optimization. To that end, we embed an optimization framework within a traditional SEIR epidemiological model that differentiates policies based on both population groups and activities, and balances the lost economic value with the cost of deaths. Since embedding complex controls into the highly non-linear, non-convex, SEIR epidemic model renders exact optimization methods not applicable, we focus on designing an algorithm that can tractably produce approximate solutions. The algorithm is a linearization and optimization procedure that we call Re-Optimization with Linearized Dynamics (ROLD), and that is inspired by model predictive control (\u003Ca id=\u0022xref-ref-7-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-7\u0022\u003EBemporad 2006\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-12-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-12\u0022\u003ECamacho and Alba 2013\u003C\/a\u003E) and trust region methods (Yuan 2015). ROLD sidesteps the lack of convexity in the original problem by repeatedly solving optimization problems that rely on a linearized approximation of the SEIR dynamical system, where each optimization problem is reduced to a linear program.\u003C\/p\u003E\u003Cp id=\u0022p-29\u0022\u003EThe second contribution of our paper is a real-life implementation of the aforementioned framework, which overcomes the data availability challenge associated with calibrating a finely targeted model like ours. Our calibration leverages publicly available data on (i) real-time hospitalization, (ii) community mobility, (iii) social contacts, and (iv) socio-economic measures collected during the COVID-19 pandemic in \u00cele-de-France, a region of France encompassing Paris with a population of approximately 12 million. Our implementation combines a complex, multi-group epidemiological model with an economic model that allows quantifying trade-offs between economic activity and epidemiological impact via the ROLD algorithm.\u003C\/p\u003E\u003Cp id=\u0022p-30\u0022\u003EWe next leverage the framework to address the core research questions that we posed earlier. We calibrate our optimization model up to October 21, 2020, and then apply our optimization framework to derive optimal intervention policies for subsequent horizons of 90 and 360 days. We draw the following conclusions from these experiments:\u003C\/p\u003E\u003Col class=\u0022list-ord \u0022 id=\u0022list-1\u0022\u003E\u003Cli id=\u0022list-item-1\u0022\u003E\u003Cp id=\u0022p-31\u0022\u003EDual targeting leads to significant Pareto improvements. Specifically, the optimal ROLD policy that differentially targets both on age groups and activities Pareto dominates the optimal ROLD policies that can only target on age groups, activities, or neither, i.e., it leads to lower economic costs without increasing the number of pandemic deaths. The gains from targeting are also super-additive: targeting on both dimensions gains more than the sum of gains from unilateral targeting. In addition, age group and activity targeted ROLD Pareto-dominates a number of benchmark policies resembling those implemented in practice during COVID-19. Finally, although not an explicit objective of the optimization, the dual-targeted ROLD policy generally results in a reduction in confinements for \u003Cem\u003Eall\u003C\/em\u003E population groups, relative to less fine-grained policies.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-2\u0022\u003E\u003Cp id=\u0022p-32\u0022\u003EOptimized dual-targeted policies have an interpretable structure, imposing less confinement on group-activity pairs that generate a relatively high economic value prorated by activity-specific social contacts. This allows ROLD to create complementary confinement schedules for different groups that reduce both the number of deaths and economic losses, with the important added benefit that they do not require completely confining any age group. To rigorously explore the interpretability of the policies produced by ROLD, we generate a training dataset by running it on an ensemble of instances, and then use CART to train trees that predict its decisions based on simple, transparent features. The resulting trees fit well and reveal that ROLD\u2019s targeting mechanism is interpretable. The key feature that explains how the targeting is done is the \u201cecon-to-contacts-ratio\u201d, i.e., the \u003Cem\u003Eratio of marginal economic value to total contacts\u003C\/em\u003E generated for a given (group, activity) pair; this provides policy makers with a transparent metric to explain ROLD\u2019s decisions.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-3\u0022\u003E\u003Cp id=\u0022p-33\u0022\u003ESince targeted policies could lead to increased disparities in the confinements of distinct groups and thus be perceived as unfair, we incorporate fairness considerations into our framework through \u201climited disparity\u201d constraints that restrict the allowed difference in confinements faced by distinct groups, and we quantify their impact on efficiency (measured in terms of health and economic costs). While the experiments suggest that the cost of limited disparity may be high, satisfactory intermediate trade-offs may be achievable through limited targeting.\u003C\/p\u003E\u003C\/li\u003E\u003C\/ol\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-7\u0022\u003E\u003Ch2 class=\u0022\u0022\u003E2. Literature Review\u003C\/h2\u003E\u003Cp id=\u0022p-34\u0022\u003EThe literature on pandemic response, particularly in the COVID-19 context, is already vast, so we focus our literature review on three key dimensions that our work most closely relates to.\u003C\/p\u003E\u003Cdiv id=\u0022sec-8\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003ETargeting\u003C\/h3\u003E\u003Cp id=\u0022p-35\u0022\u003EParalleling our aforementioned real-life examples, several papers have studied targeted interventions. \u003Ca id=\u0022xref-ref-31-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-31\u0022\u003EKucharski et al. (2020\u003C\/a\u003E), \u003Ca id=\u0022xref-ref-41-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-41\u0022\u003EPrem et al. (2020\u003C\/a\u003E), \u003Ca id=\u0022xref-ref-15-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-15\u0022\u003EDi Domenico et al. (2020)\u003C\/a\u003E recognize the importance of heterogeneity in the social contacts generated through activities and examine several interventions limiting them. Though some of the models here are age differentiated, targeting only happens through activities. Population group targeting, either through confinements, testing or vaccinations, has been investigated in \u003Ca id=\u0022xref-ref-6-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-6\u0022\u003EBastani et al. (2021)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-1-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-1\u0022\u003EAcemoglu et al. (2020)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-36-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-36\u0022\u003EMatrajt et al. (2020)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-26-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-26\u0022\u003EGoldstein et al. (2021)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-9-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-9\u0022\u003EBertsimas et al. (2020)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-21-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-21\u0022\u003EFavero et al. (2020)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-10-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-10\u0022\u003EBirge et al. (2020)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-13-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-13\u0022\u003EChang et al. (2020)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-20-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-20\u0022\u003EEvgeniou et al. (2020)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-24-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-24\u0022\u003EGiordano et al. (2021)\u003C\/a\u003E. By enforcing stricter confinements for higher risk groups (e.g., older populations when considering mortality risk or younger populations when considering the risk of new infections), such targeted policies have been shown to generate potentially significant improvements in health outcomes, and even in economic value if optimally tailored (\u003Ca id=\u0022xref-ref-1-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-1\u0022\u003EAcemoglu et al. 2020\u003C\/a\u003E). A potential benefit of our approach is that, by exploiting complementarities between group and activity targeting, these higher risk groups may experience less confinement for the same level of aggregate deaths and economic losses.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-9\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003EOptimization of interventions\u003C\/h3\u003E\u003Cp id=\u0022p-36\u0022\u003EOur proposed optimization algorithm differs in several ways from existing approaches to assessing interventions via SEIR epidemiological models. A number of papers simulate a small number of candidate interventions, e.g. full lockdown versus school-only lockdown (\u003Ca id=\u0022xref-ref-31-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-31\u0022\u003EKucharski et al. 2020\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-41-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-41\u0022\u003EPrem et al. 2020\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-15-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-15\u0022\u003EDi Domenico et al. 2020\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-19-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-19\u0022\u003EEl Housni et al. 2020\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-21-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-21\u0022\u003EFavero et al. 2020\u003C\/a\u003E), restrict the candidates to a simple parametric class for which exhaustive search is computationally feasible (e.g., trigger policies based on hospital admissions as in \u003Ca id=\u0022xref-ref-17-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-17\u0022\u003EDuque et al. 2020\u003C\/a\u003E or confirmed cases as \u003Ca id=\u0022xref-ref-3-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-3\u0022\u003EAhn et al. 2021\u003C\/a\u003E), or use global optimization methods such as simulated annealing (\u003Ca id=\u0022xref-ref-18-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-18\u0022\u003EDutta et al. 2021\u003C\/a\u003E). When considering a more complex policy space like in our targeting model, such approaches can lead to significantly sub-optimal results and misleading conclusions. Another stream borrows from the optimal control literature to design non-targeted control policies for single-group and single-activity SEIR dynamics (\u003Ca id=\u0022xref-ref-11-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-11\u0022\u003EBose et al. 2021\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-38-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-38\u0022\u003EPataro et al. 2021\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-37-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-37\u0022\u003EMorris et al. 2021\u003C\/a\u003E). Although optimization-based, the models there are simpler than our own and do not capture targeting. \u003Ca id=\u0022xref-ref-10-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-10\u0022\u003EBirge et al. (2020)\u003C\/a\u003E use formal optimization for location-based targeting, but in a one-shot model that does not differentiate age groups or activities and does not account for time in the calculation of health or economic impact. The paper that is most related to ours is \u003Ca id=\u0022xref-ref-9-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-9\u0022\u003EBertsimas et al. (2020)\u003C\/a\u003E, which also proposes a multi-group SEIR formulation, in conjunction with an iterative coordinate descent algorithm to optimize vaccine allocations in a differentiated fashion. Although taking different approaches to doing so, both the algorithm proposed there and the ROLD heuristic crucially depend on solving linearized versions of the true SEIR dynamics which are tractable via commercial solvers. However, the model of \u003Ca id=\u0022xref-ref-9-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-9\u0022\u003EBertsimas et al. (2020)\u003C\/a\u003E focuses on vaccine allocation decisions, whereas ours captures the dynamics of differential confinements and also allows activity-based targeting.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-10\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003ESEIR model calibration\u003C\/h3\u003E\u003Cp id=\u0022p-37\u0022\u003EOur work is also related to several other papers that have estimated SEIR parameters, particularly in the COVID-19 context. A number of papers estimate SEIR epidemiological parameters stratified by age groups \u2013 we use the estimates from the \u00cele-de-France study in \u003Ca id=\u0022xref-ref-44-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. (2020\u003C\/a\u003E). Another stream of papers, focusing on forecasting COVID-19 spread, estimate when and how the underlying SEIR parameters evolve in response to government interventions and changes in individual behavior, such as \u003Ca id=\u0022xref-ref-39-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-39\u0022\u003EPerakis et al. (2021)\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-33-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-33\u0022\u003ELi et al. (2021)\u003C\/a\u003E. Lastly, we relate to the literature that has used Google and other mobility data to inform COVID-19 response strategies or to estimate the realized reductions in social contacts during COVID-19 (\u003Ca id=\u0022xref-ref-18-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-18\u0022\u003EDutta et al. 2021\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-29-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-29\u0022\u003EIlin et al. 2021\u003C\/a\u003E, Wellenius et al. 2020, \u003Ca id=\u0022xref-ref-14-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-14\u0022\u003ECot et al. 2021\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-49-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-49\u0022\u003EXiong et al. 2020\u003C\/a\u003E), as well as the literature on social contacts estimation (\u003Ca id=\u0022xref-ref-8-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-8\u0022\u003EB\u00e9raud et al. 2015\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-40-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-40\u0022\u003EPrem et al. 2017\u003C\/a\u003E).\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-11\u0022\u003E\u003Ch2 class=\u0022\u0022\u003E3. Model and Optimization Problem\u003C\/h2\u003E\u003Cp id=\u0022p-38\u0022\u003EOur framework relies on a flexible model that captures several important real-world considerations. We extend a multi-group SEIR model to capture controls that target based on (i) \u003Cem\u003Eage groups\u003C\/em\u003E, and (ii) types of \u003Cem\u003Eactivities\u003C\/em\u003E that individuals engage in. Different policy interventions can be embedded: we include time-dependent confinements as well as testing and quarantining in our study, but vaccinations can also be accommodated. The controls modulate the rate of social contacts and the economic value generated, and the objective of the control problem is to minimize a combination of health and economic losses caused by deaths, illness, and activity restrictions. The model captures important resource constraints (such as hospital, ICU, and testing capacities), and allows explicitly controlling the amount of targeting through \u201climited disparity\u201d constraints that limit the difference in the extent of confinement imposed on distinct population groups.\u003C\/p\u003E\u003Cdiv id=\u0022sec-12\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E3.1. Some Notation\u003C\/h3\u003E\u003Cp id=\u0022p-39\u0022\u003EWe denote scalars by lower-case letters, as in \u003Cem\u003Ev\u003C\/em\u003E, and vectors by bold letters, as in \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E. We use square brackets to denote the concatenation into vectors: \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E := [\u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E0\u003C\/sub\u003E, \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E]. For a time series of vectors \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E1\u003C\/sub\u003E, \u2026, \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003En\u003C\/em\u003E\u003C\/sub\u003E, we use the notation \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ei\u003C\/em\u003E:\u003Cem\u003Ej\u003C\/em\u003E\u003C\/sub\u003E := [\u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ei\u003C\/em\u003E\u003C\/sub\u003E, \u2026, \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ej\u003C\/em\u003E\u003C\/sub\u003E] to denote the concatenation of vectors \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ei\u003C\/em\u003E\u003C\/sub\u003E through \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ej\u003C\/em\u003E\u003C\/sub\u003E. Lastly, we use \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E\u003Csup\u003E\u22a4\u003C\/sup\u003E to refer to the transpose of \u003Cstrong\u003E\u003Cem\u003Ev\u003C\/em\u003E\u003C\/strong\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-13\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E3.2. Epidemiological Model and Controls\u003C\/h3\u003E\u003Cp id=\u0022p-40\u0022\u003EWe rely on a modified version of the discretized SEIR (Susceptible-Exposed-Infectious-Recovered) epidemiological model (\u003Ca id=\u0022xref-ref-5-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-5\u0022\u003EAnderson and May 1992\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-41-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-41\u0022\u003EPrem et al. 2020\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-44-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. 2020\u003C\/a\u003E) with multiple population groups that interact with each other. In our case study we use nine groups \u003Cem\u003Eg\u003C\/em\u003E \u2208 \ud835\udca2 determined by age and split in 10-year buckets, with the youngest group capturing individuals with age 0-9 and the oldest capturing individuals with age 80 or above. Time is discrete, indexed by \u003Cem\u003Et\u003C\/em\u003E = 0, 1, \u2026, \u003Cem\u003ET\u003C\/em\u003E and measured in days. We assume that no infections are possible beyond time \u003Cem\u003ET\u003C\/em\u003E.\u003C\/p\u003E\u003Cdiv id=\u0022sec-14\u0022 class=\u0022subsection\u0022\u003E\u003Ch4\u003ECompartmental Model and States\u003C\/h4\u003E\u003Cp id=\u0022p-41\u0022\u003E\u003Ca id=\u0022xref-fig-1-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F1\u0022\u003EFigure 1\u003C\/a\u003E represents the compartmental model and the SEIR transitions for a specific group \u003Cem\u003Eg\u003C\/em\u003E. For a population group \u003Cem\u003Eg\u003C\/em\u003E in time period \u003Cem\u003Et\u003C\/em\u003E, the compartmental model includes states \u003Cem\u003ES\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) (susceptible to be infected), \u003Cem\u003EE\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) (exposed but not yet infectious), \u003Cem\u003EI\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) (infectious but not confirmed through testing and thus not quarantined), \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-1\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-1.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-1.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E (infectious and confirmed through testing and thus quarantined; this state is further subdivided into \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-2\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-2.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-2.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E for \u003Cem\u003Ej\u003C\/em\u003E \u2208 {\u003Cem\u003Ea, ps, ms, ss\u003C\/em\u003E} to model different degrees of severity of symptoms: asymptomatic, paucisymptomatic, with mild symptoms, or with severe symptoms; we assume that an infectious individual in group \u003Cem\u003Eg\u003C\/em\u003E will exhibit symptoms of degree \u003Cem\u003Ej\u003C\/em\u003E with probability \u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Ej,g\u003C\/em\u003E\u003C\/sub\u003E), \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) (recovered but not confirmed as having had the virus), \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-3\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-3.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-3.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E (recovered and confirmed as having had the virus), and \u003Cem\u003ED\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) (deceased). Individuals with severe symptoms will need hospitalization, either in general hospital wards (\u003Cem\u003EH\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E)) or in intensive care units (\u003Cem\u003EICU\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E)). All the states represent the number of individuals in a compartment of the model in the \u003Cem\u003Ebeginning\u003C\/em\u003E of the time period.\u003C\/p\u003E\u003Cdiv id=\u0022F1\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F1.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Compartmental SEIR model for a specific group g with transition rates. \u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;\u0026lt;span class=\u0026quot;caption-title\u0026quot;\u0026gt;Compartmental SEIR model for a specific group \u0026lt;em\u0026gt;g\u0026lt;\/em\u0026gt; with transition rates.\u0026lt;\/span\u0026gt; \u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure 1\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F1.medium.gif\u0022 width=\u0022440\u0022 height=\u0022207\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure 1\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F1.medium.gif\u0022 width=\u0022440\u0022 height=\u0022207\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F1.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure 1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F1.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022 xmlns:xhtml=\u0022http:\/\/www.w3.org\/1999\/xhtml\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure 1\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003ECompartmental SEIR model for a specific group \u003Cem\u003Eg\u003C\/em\u003E with transition rates.\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-42\u0022\u003ESusceptible individuals get infected and transition to the exposed state at a rate determined by the number of social contacts and the transmission rate \u003Cem\u003E\u03b2\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E). Exposed individuals transition to the infectious state at a rate \u003Cem\u003E\u03c3\u003C\/em\u003E and infectious individuals transition out of the infectious state at a rate \u003Cem\u003E\u00b5\u003C\/em\u003E. An infectious individual will need to be hospitalized in general hospital wards or ICU with probability \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-4\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-4.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-4.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-5\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-5.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-5.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, respectively, where \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-6\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-6.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-6.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. On average, patients who are treated in general hospital wards (ICU) spend \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-7\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-7.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-7.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E days in the hospital (ICU). An infectious individual with severe symptoms in group \u003Cem\u003Eg\u003C\/em\u003E will decease (recover) with probability \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-8\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-8.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-8.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-43\u0022\u003EWe keep track of all living individuals in group \u003Cem\u003Eg\u003C\/em\u003E who are not confirmed to have had the disease \u003Cem\u003EN\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) := \u003Cem\u003ES\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003EE\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003EI\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E), and let \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E = \u003Cem\u003ES\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003EI\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u2026, \u003Cem\u003ED\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E)] \u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E denote the full state of the system (across groups) at time 0\u2264 \u003Cem\u003Et\u003C\/em\u003E\u2264 \u003Cem\u003ET\u003C\/em\u003E. We denote the number of compartments by |\ud835\udcb3|, so the dimension of \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E is |\ud835\udca2||\ud835\udcb3| \u00d71.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-15\u0022 class=\u0022subsection\u0022\u003E\u003Ch4\u003EControls\u003C\/h4\u003E\u003Cp id=\u0022p-44\u0022\u003EIndividuals interact in activities belonging to the set \ud835\udc9c = {\u003Ckbd\u003Ework\u003C\/kbd\u003E, \u003Ckbd\u003Etransport\u003C\/kbd\u003E, \u003Ckbd\u003Eleisure\u003C\/kbd\u003E, \u003Ckbd\u003Eschool\u003C\/kbd\u003E, \u003Ckbd\u003Ehome\u003C\/kbd\u003E, \u003Ckbd\u003Eother\u003C\/kbd\u003E }. These interactions generate \u003Cem\u003Esocial contacts\u003C\/em\u003E that drive the rate of new infections.\u003C\/p\u003E\u003Cp id=\u0022p-45\u0022\u003EWe control the SEIR dynamics by adjusting the confinement intensity in each group-activity pair over time: we let \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-9\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-9.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-9.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E denote the activity level allowed for group \u003Cem\u003Eg\u003C\/em\u003E and activity \u003Cem\u003Ea\u003C\/em\u003E at time \u003Cem\u003Et\u003C\/em\u003E, expressed as a fraction of the activity level under a \u003Cem\u003Enormal course\u003C\/em\u003E of life (i.e., no confinement). In our study we take \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-10\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-10.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-10.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, meaning that the number of social contacts at home is unchanged irrespective of confinement policy.\u003Csup\u003E\u003Ca id=\u0022xref-fn-6-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-6\u0022\u003E1\u003C\/a\u003E\u003C\/sup\u003E We denote the vector of all activity levels for group \u003Cem\u003Eg\u003C\/em\u003E at \u003Cem\u003Et\u003C\/em\u003E by \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-11\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-11.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-11.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, and we also refer to \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) as confinement decisions when no confusion can arise.\u003C\/p\u003E\u003Cp id=\u0022p-46\u0022\u003EWe propose a parametric model to map activity levels to social contacts. We use \u003Cem\u003Ec\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg,h\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eh\u003C\/em\u003E\u003C\/sub\u003E) to denote the mean number of total daily contacts between an individual in group \u003Cem\u003Eg\u003C\/em\u003E and individuals in group \u003Cem\u003Eh\u003C\/em\u003E across all activities when their activity levels are \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eh\u003C\/em\u003E\u003C\/sub\u003E, respectively. Varying the activity levels changes the social contacts according to\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-1\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-2.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-2.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-12\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-12.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-12.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E denote the mean number of daily contacts in activity \u003Cem\u003Ea\u003C\/em\u003E under normal course (i.e., without confinement), and \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E, \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E2\u003C\/sub\u003E \u2208 \u211d are parameters. This parametrization is similar to a Cobb-Douglas production function (\u003Ca id=\u0022xref-ref-35-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-35\u0022\u003EMas-Colell et al. 1995\u003C\/a\u003E), using the confinement patterns as inputs, and the number of social contacts as output. We retrieve values for \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-13\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-13.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-13.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E from the data tool of \u003Ca id=\u0022xref-ref-48-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-48\u0022\u003EWille et al. (2020)\u003C\/a\u003E, which is based on the French social contact survey data in \u003Ca id=\u0022xref-ref-8-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-8\u0022\u003EB\u00e9raud et al. (2015\u003C\/a\u003E), and we estimate \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E, \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E2\u003C\/sub\u003E from health outcome data (\u003Ca id=\u0022xref-ref-23-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-23\u0022\u003EFrench Government 2020\u003C\/a\u003E) and Google mobility data (\u003Ca id=\u0022xref-ref-27-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-27\u0022\u003EGoogle 2020\u003C\/a\u003E), as described in \u003Ca id=\u0022xref-sec-50-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-50\u0022\u003ESection EC.4.2\u003C\/a\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-47\u0022\u003EBesides confinements, we also model targeted viral testing decisions, which capture how much of a finite capacity of tests to allocate to each age group. We model random mass testing, where a test detects an infectious individual with probability equal to the fraction of infectious individuals in the group\u2019s population. Infected individuals in group \u003Cem\u003Eg\u003C\/em\u003E who are detected are placed in the quarantined SEIR state \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-14\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-14.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-14.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, where they can no longer infect others. We use \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-15\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-15.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-15.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E to denote the number of viral tests allocated to group \u003Cem\u003Eg\u003C\/em\u003E in period \u003Cem\u003Et\u003C\/em\u003E. The testing decisions for the policy maker are then \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-16\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-16.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-16.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. \u003Ca id=\u0022xref-fig-1-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F1\u0022\u003EFigure 1\u003C\/a\u003E represents the flows from one compartment to the other resulting from testing, for a specific group \u003Cem\u003Eg\u003C\/em\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-48\u0022\u003ELet\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-2\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-3.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-3.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\ndenote the vector of all decisions at time \u003Cem\u003Et\u003C\/em\u003E\u2208 {0, 1, \u2026, \u003Cem\u003ET\u003C\/em\u003E \u22121 }, i.e., the confinement and viral testing decisions for all the groups. We denote the number of different decisions for a given group at a given time by |\ud835\udcb0|. Then the dimension of \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E is |\ud835\udca2||\ud835\udcb0| \u00d7 1.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-16\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E3.3. Resources and Constraints\u003C\/h3\u003E\u003Cp id=\u0022p-49\u0022\u003EWe use \u003Cem\u003EK\u003C\/em\u003E\u003Csup\u003EH\u003C\/sup\u003E(\u003Cem\u003Et\u003C\/em\u003E) (\u003Cem\u003EK\u003C\/em\u003E\u003Csup\u003EICU\u003C\/sup\u003E(\u003Cem\u003Et\u003C\/em\u003E)) to denote the capacity of beds in the general hospital wards (in the ICU) on day \u003Cem\u003Et\u003C\/em\u003E. When the patient inflow into the hospital or the ICU exceeds the remaining number of available beds, then the policy maker needs to decide how many patients to turn away from each group. Although in principle our framework allows optimizing over such decisions, we choose to not consider this dimension of targeting because it can be extremely contentious in practice. Instead, we implement a proportional rule that allocates any remaining hospital and ICU capacity among patients from all age groups proportionally to the number of cases requiring admission from each group. More formally, with \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-17\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-17.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-17.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-18\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-18.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-18.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E denoting the number of patients from age group \u003Cem\u003Eg\u003C\/em\u003E who are denied admission to general hospital wards and ICU, respectively, in period \u003Cem\u003Et\u003C\/em\u003E, the proportional rule is:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-3\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-4.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-4.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-4\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-5.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-5.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-19\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-19.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-19.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003Eand \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-20\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-20.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-20.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E denote the inflow of patients from group \u003Cem\u003Eg\u003C\/em\u003E into the hospital and ICU in period \u003Cem\u003Et\u003C\/em\u003E, respectively. We assume that all patients that are denied admission die immediately.\u003C\/p\u003E\u003Cp id=\u0022p-50\u0022\u003EWe further assume a given capacity for viral tests each day, which we denote by \u003Cem\u003EK\u003C\/em\u003E\u003Csup\u003EVtest\u003C\/sup\u003E(\u003Cem\u003Et\u003C\/em\u003E) on day \u003Cem\u003Et\u003C\/em\u003E. We assume that viral tests used to test individuals with severe symptoms that enter the ICU or hospital, as well as viral tests that test hospitalized individuals to confirm they have recovered, come from a different pool of tests and do not consume the capacity for viral testing in the non-hospitalized population.\u003C\/p\u003E\u003Cp id=\u0022p-51\u0022\u003EWe can now write a complete set of discrete dynamical equations for the controlled SEIR model ((EC.1)-(EC.11) in Appendix EC.1) and summarize these using the function\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-5\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-6.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-6.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u0394\u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E := \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E+1\u003C\/sub\u003E \u2212 \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E. Additionally, we can also write the following constraints for the optimization problem:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-6\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-7.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-7.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-7\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-8.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-8.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-8\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-9.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-9.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-9\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-10.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-10.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-10\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-11.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-11.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nWe denote by \ud835\udc9e (\u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E) the feasible set described by (5)-(9) for the vector of decisions \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E at time \u003Cem\u003Et\u003C\/em\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-17\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E3.4. Objective\u003C\/h3\u003E\u003Cp id=\u0022p-52\u0022\u003EOur objective captures two criteria. The first quantifies the total deaths directly attributable to the pandemic, which we denote by \u003Ckbd\u003ETotal Deaths\u003C\/kbd\u003E(\u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E0:\u003Cem\u003ET\u003C\/em\u003E \u22121\u003C\/sub\u003E) := \u2211\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E \u003Cem\u003ED\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003ET\u003C\/em\u003E) to reflect the dependency on the specific policy \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E0:\u003Cem\u003ET\u003C\/em\u003E \u22121\u003C\/sub\u003E followed.\u003C\/p\u003E\u003Cp id=\u0022p-53\u0022\u003EThe second criterion captures the economic losses due to the pandemic, denoted by \u003Ckbd\u003EEconomic Loss\u003C\/kbd\u003E(\u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E0:\u003Cem\u003ET\u003C\/em\u003E \u22121\u003C\/sub\u003E). These stem from three sources: (a) lost productivity due to confinement, (b) lost productivity during the pandemic due to individuals being quarantined, hospitalized, or deceased, and (c) lost value after the pandemic due to deaths (as deceased individuals no longer produce economic output even after the pandemic ends).\u003C\/p\u003E\u003Cp id=\u0022p-54\u0022\u003ETo model (a), we assign a daily economic value \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E) to each individual in group \u003Cem\u003Eg\u003C\/em\u003E that depends on the activity levels \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E := [\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E]\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E across all groups and activities. For the working age groups, \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E) comes from wages from employment and is a linear function of group \u003Cem\u003Eg\u003C\/em\u003E\u2019s activity level in \u003Ckbd\u003Ework\u003C\/kbd\u003E \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-21\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-21.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-21.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and of the average activity levels in \u003Ckbd\u003Eleisure\u003C\/kbd\u003E, \u003Ckbd\u003Eother\u003C\/kbd\u003E and \u003Ckbd\u003Etransport\u003C\/kbd\u003E for the entire population (equally weighted). This reflects that the value generated in some industries, like retail, is impacted by confinements across all these three activities. For the school age groups, \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E) captures future wages from employment due to schooling and depends only on the group\u2019s activity level in \u003Ckbd\u003Eschool\u003C\/kbd\u003E \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-22\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-22.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-22.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. For (b), we assume that an individual who is in quarantine, hospitalized, or deceased, generates no economic value. For (c), we determine the wages that a deceased individual would have earned based on their current age until retirement age under the prevailing wage curve, and denote the resulting amount of lost wages with \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-23\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-23.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-23.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-55\u0022\u003EThe overall economic loss is the difference between the economic value that would have been generated during the time of the pandemic under a \u201cno pandemic\u201d scenario, denoted by \u003Cem\u003EV\u003C\/em\u003E, and the value generated during the pandemic, plus the future economic output lost due to deaths.\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-11\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-12.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-12.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nAll the details of the economic modelling are deferred to Appendix EC.2.\u003C\/p\u003E\u003Cp id=\u0022p-56\u0022\u003ETo allow policy makers to weigh the importance of the two criteria, we associate a cost \u003Cem\u003E\u03c7\u003C\/em\u003E to each death, which we express in multiples of French GDP per capita. Our framework can capture a multitude of policy preferences by considering a wide range of \u003Cem\u003E\u03c7\u003C\/em\u003E values, from completely prioritizing economic losses (\u003Cem\u003E\u03c7\u003C\/em\u003E = 0) to completely prioritizing deaths (\u003Cem\u003E\u03c7\u003C\/em\u003E \u2192 \u221e).\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-18\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E3.5. Optimization Problem\u003C\/h3\u003E\u003Cp id=\u0022p-57\u0022\u003EThe optimization problem we solve is to find control policies (for confinement and testing) that minimize the sum of mortality and economic losses\u003Csup\u003E\u003Ca id=\u0022xref-fn-7-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-7\u0022\u003E2\u003C\/a\u003E\u003C\/sup\u003E subject to the constraints that (i) the state trajectory follows the SEIR dynamics, and (ii) the controls and states respect the capacity and feasibility constraints discussed above. Formally, we solve:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-12\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-13.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-13.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-13\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-14.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-14.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-14\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-15.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-15.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-19\u0022\u003E\u003Ch2 class=\u0022\u0022\u003E4. Algorithm: Re-Optimization with Linearized Dynamics\u003C\/h2\u003E\u003Cp id=\u0022p-58\u0022\u003EBefore describing our approach to solving problem (11)-(13), we first comment briefly on the challenges in solving this problem to optimality, i.e. without resorting to approximation methods. A key term in the dynamics of any SEIR-type model is the rate of new infections, which involves multiplying the current susceptible population in a given group with the infected population in another group. This introduces non-linearity in the state trajectory; for instance, our dynamic for the evolution of the susceptible population in group \u003Cem\u003Eg\u003C\/em\u003E from (EC.2) reads:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-15\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-16.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-16.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nIt can be easily seen that expanding out \u003Cem\u003ES\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) produces a nested fraction of polynomials in all the past decisions \u003Cem\u003El\u003C\/em\u003E(\u003Cem\u003E\u03c4\u003C\/em\u003E), \u003Cem\u003EN\u003C\/em\u003E \u003Csup\u003EVtest\u003C\/sup\u003E(\u003Cem\u003E\u03c4\u003C\/em\u003E) for 0 \u2264 \u003Cem\u003E\u03c4\u003C\/em\u003E \u2264 \u003Cem\u003Et\u003C\/em\u003E\u2212 1.\u003Csup\u003E\u003Ca id=\u0022xref-fn-8-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-8\u0022\u003E3\u003C\/a\u003E\u003C\/sup\u003E This function has no identifiable structure that would make the resulting optimization problem tractable via convex optimization. Similarly, the objective also involves products of states and controls, and suffers from the same lack of convexity.\u003C\/p\u003E\u003Cp id=\u0022p-59\u0022\u003EWith this in mind, we focus on developing heuristics that can tractably yield good policies, and propose an algorithm, which we call Re-Optimization with Linearized Dynamics, or ROLD, that builds a control policy by incrementally solving linear approximations of the true SEIR system.\u003C\/p\u003E\u003Cdiv id=\u0022sec-20\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E4.1. Linearization and Optimization\u003C\/h3\u003E\u003Cp id=\u0022p-60\u0022\u003EThe key idea is to solve the problem in a shrinking-horizon fashion, where at each time step \u003Cem\u003Ek\u003C\/em\u003E = 0, \u2026, \u003Cem\u003ET\u003C\/em\u003E we linearize the system dynamics and objective (over the remaining horizon), determine optimal decisions for all \u003Cem\u003Ek\u003C\/em\u003E, \u2026, \u003Cem\u003ET\u003C\/em\u003E, and only implement the decisions for the current time step \u003Cem\u003Ek\u003C\/em\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-61\u0022\u003EWe first describe the linearization procedure. Recall that the true evolution of our dynamical system is given by (4). The typical approach in dynamical systems is to linearize the system dynamics around a particular \u201cnominal\u201d trajectory. More precisely, assume that at time \u003Cem\u003Ek\u003C\/em\u003E we have access to a nominal control sequence \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-24\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-24.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-24.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and let \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-25\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-25.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-25.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E denote the resulting nominal system trajectory under the true dynamic (4) and under \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-26\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-26.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-26.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. We approximate the original dynamics through a Taylor expansion around \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-27\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-27.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-27.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-16\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-17.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-17.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u2207\u003Csub\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/sub\u003E\u003Cstrong\u003E\u003Cem\u003EF\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E and \u2207\u003Csub\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/sub\u003E\u003Cstrong\u003E\u003Cem\u003EF\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E denote the Jacobians with respect to \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E and \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E, respectively. Note that these Jacobians are evaluated at points on the nominal trajectory, so (14) is indeed a linear expression of \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E and \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E. By induction, every state \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E under dynamic (14) will be a linear function of \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003E\u03c4\u003C\/em\u003E\u003C\/sub\u003E for \u003Cem\u003E\u03c4 \u0026lt; t\u003C\/em\u003E, and all the constraints will also depend linearly on the decisions.\u003C\/p\u003E\u003Cp id=\u0022p-62\u0022\u003EIn a similar fashion, we also linearize the objective (11). Since \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E(\u003Cem\u003Et\u003C\/em\u003E)) is linear in \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E for all \u003Cem\u003Et\u003C\/em\u003E = 0, \u2026, \u003Cem\u003ET\u003C\/em\u003E \u2212 1, the objective contains \u003Cem\u003Ebilinear\u003C\/em\u003E terms and can be written compactly as:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-17\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-18.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-18.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nfor some matrix \u003Cstrong\u003E\u003Cem\u003EM\u003C\/em\u003E\u003C\/strong\u003E with dimensions |\ud835\udca2||\ud835\udcb0| \u00d7 |\ud835\udca2||\ud835\udcb3|, and vectors \u003Cstrong\u003E\u003Cem\u003E\u03b3\u003C\/em\u003E\u003C\/strong\u003E and \u003Cstrong\u003E\u003Cem\u003E\u03b7\u003C\/em\u003E\u003C\/strong\u003E of dimensions |\ud835\udca2||\ud835\udcb3| \u00d71 (detailed expressions are available in Appendix EC.3). By linearizing this using a Taylor approximation, we consider the following objective instead:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-18\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-19.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-19.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhich depends linearly on all the decisions \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E0\u003C\/sub\u003E,\u2026, \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003ET\u003C\/em\u003E \u22121\u003C\/sub\u003E.\u003C\/p\u003E\u003Cdiv id=\u0022sec-21\u0022 class=\u0022subsection\u0022\u003E\u003Ch4\u003ELinearization-optimization procedure\u003C\/h4\u003E\u003Cp id=\u0022p-63\u0022\u003EWe use the following heuristic to obtain an approximate control at time \u003Cem\u003Ek\u003C\/em\u003E, for \u003Cem\u003Ek\u003C\/em\u003E = 0,\u003Cem\u003E\u2026, T\u003C\/em\u003E \u2212 1:\u003C\/p\u003E\u003Col class=\u0022list-ord \u0022 id=\u0022list-2\u0022\u003E\u003Cli id=\u0022list-item-4\u0022\u003E\u003Cp id=\u0022p-64\u0022\u003EGiven the current state \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ek\u003C\/em\u003E\u003C\/sub\u003E and a nominal control sequence \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-28\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-28.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-28.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E for all remaining periods, calculate a nominal system trajectory \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-29\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-29.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-29.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E under the true dynamic in (4). (The nominal control sequence is set to a solution obtained by a gradient descent method at \u003Cem\u003Ek\u003C\/em\u003E = 0, and to the algorithm\u2019s own output from period \u003Cem\u003Ek\u003C\/em\u003E\u2212 1 for periods \u003Cem\u003Ek \u0026gt;\u003C\/em\u003E 0, per Step 4 below.)\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-5\u0022\u003E\u003Cp id=\u0022p-65\u0022\u003EUse (14) to approximate the state dynamic around the nominal trajectory \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-30\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-30.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-30.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and use (16) to approximate the objective-to-go function over the remaining periods \u003Cem\u003Et\u003C\/em\u003E \u2208 {\u003Cem\u003Ek\u003C\/em\u003E, \u2026, \u003Cem\u003ET\u003C\/em\u003E }.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-6\u0022\u003E\u003Cp id=\u0022p-66\u0022\u003ESolve the linear program to obtain decision variables \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-31\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-31.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-31.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E that maximize the linearized objective-to-go subject to all the relevant linearized constraints.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-7\u0022\u003E\u003Cp id=\u0022p-67\u0022\u003ESet the nominal control sequence for the next time step as \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-32\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-32.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-32.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-8\u0022\u003E\u003Cp id=\u0022p-68\u0022\u003EUpdate the states using the optimal control \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-33\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-33.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-33.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and the true dynamic in (4), i.e.\u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-34\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-34.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-34.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003C\/li\u003E\u003C\/ol\u003E\u003Cp id=\u0022p-69\u0022\u003EThe linearization-optimization procedure described above is run for all periods \u003Cem\u003Ek\u003C\/em\u003E = 0, \u2026, \u003Cem\u003ET\u003C\/em\u003E \u2212 1 sequentially to output a full control policy\u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-35\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-35.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-35.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-22\u0022 class=\u0022subsection\u0022\u003E\u003Ch4\u003ETrust region implementation\u003C\/h4\u003E\u003Cp id=\u0022p-70\u0022\u003EIn our experiments, we have found that the linearized model described in (14) may diverge significantly from the real dynamical system when the optimized controls \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-36\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-36.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-36.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E determined in Step 3 diverge sufficiently from the nominal controls \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-37\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-37.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-37.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E considered in the linearization in Step 2. This can lead to a large sensitivity in performance to the initialization used in the very first step; for example, if the Taylor approximation were constructed around a policy of full confinement, the linearized model could systematically underestimate the number of infections and deaths created when considering more relaxed confinements.\u003C\/p\u003E\u003Cp id=\u0022p-71\u0022\u003EWe overcome this by employing an iterative procedure inspired by a trust region optimization method. The key idea is to avoid the large approximation errors by running the linearization-optimization procedure iteratively within each time step \u003Cem\u003Ek\u003C\/em\u003E, with each iteration only being allowed to take a small step towards the optimum within a trust region of an \u003Cem\u003E\u03f5\u003C\/em\u003E-ball around the nominal control sequence \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-38\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-38.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-38.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, and the updated optimized control sequence of each iteration being used as a nominal sequence for the next iteration. This leads to a procedure that is much more robust to the initial guess of control sequence, albeit at the expense of increased computation time.\u003C\/p\u003E\u003Cp id=\u0022p-72\u0022\u003EFurther algorithmic details for ROLD are provided in Appendix EC.3.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-23\u0022\u003E\u003Ch2 class=\u0022\u0022\u003E5. \u00cele-de-France Model Parametrization, Calibration and Experimental Setup\u003C\/h2\u003E\u003Cp id=\u0022p-73\u0022\u003EIn this section we summarize our approach for parameter specification and model calibration, and set up the experiments for our implementation calibrated on \u00cele-de-France COVID-19 data.\u003C\/p\u003E\u003Cdiv id=\u0022sec-24\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E5.1. Parametrization and Calibration\u003C\/h3\u003E\u003Cp id=\u0022p-74\u0022\u003EWe parametrize the epidemiological model using the confidence regions for SEIR parameters reported in \u003Ca id=\u0022xref-ref-44-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. (2020\u003C\/a\u003E) for the \u00cele-de-France region. We calibrate the model using public data on (i) community mobility, (ii) social contacts, (iii) health outcomes, and (iv) economic output. In particular, we use Google mobility data (\u003Ca id=\u0022xref-ref-27-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-27\u0022\u003EGoogle 2020\u003C\/a\u003E) to approximate the mean effective lockdowns for all activities during the horizon of interest. Based on these, as well as social contacts data for France (\u003Ca id=\u0022xref-ref-8-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-8\u0022\u003EB\u00e9raud et al. 2015\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-48-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-48\u0022\u003EWille et al. 2020\u003C\/a\u003E), we simulate our SEIR model to generate several potential sample paths, which we use in conjunction with real data from the French Public Health Agency (\u003Ca id=\u0022xref-ref-23-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-23\u0022\u003EFrench Government 2020\u003C\/a\u003E) on hospital and ICU utilization and deaths in \u00cele-de-France to generate a fitting error metric. Lastly, we estimate values of all parameters of interest by minimizing the sample-average-approximation of the error metric. We calibrate our economic model using data for France, and where available \u00cele-de-France, on full time equivalent wages and employment rates from the French National Institute of Statistics and Economic Studies, and sentiment surveys on business activity levels during confinement from the Bank of France. We provide all the details for calibration and parameter specification in Appendix EC.4. We report experimental results from sensitivity and robustness analyses on the fitted parameters in Appendix EC.6.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-25\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E5.2. Experimental and Optimization Setup\u003C\/h3\u003E\u003Cp id=\u0022p-75\u0022\u003EWe run experiments over a range of values for the parameters of our model. \u003Ca id=\u0022xref-table-wrap-1-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T1\u0022\u003ETable 1\u003C\/a\u003E summarizes the values we use for each parameter in our experimental setup, as well as the details of the optimization setup. For parameters for which multiple values are used in our experiments, the \u201cBaseline Value\u201d column reports the values of the parameters used in our baseline setting, as reported in results in the main paper and the Appendix, unless specified otherwise. In particular, we use a baseline capacity of 2900 beds for ICU in \u00cele-de-France, and also experiment with ICU capacities that range from 2000 to 3200 beds.\u003Csup\u003E\u003Ca id=\u0022xref-fn-9-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-9\u0022\u003E4\u003C\/a\u003E\u003C\/sup\u003E We use an infinite capacity for general hospital wards. All results reported in the main paper are obtained under a testing capacity of zero, so only confinement decisions are optimized and compared. (Appendix EC.6.5 discusses the additional benefits of targeted administration of viral tests across age groups.) We optimize decisions starting on October 21 2020, using an optimization horizon of \u003Cem\u003ET\u003C\/em\u003E = 90 days in the experiments reported in the main paper, and allowing up to \u003Cem\u003ET\u003C\/em\u003E = 360 days in additional experiments reported in Appendix EC.6.4. We allow the confinement decisions to change every two weeks.\u003C\/p\u003E\u003Cdiv id=\u0022T1\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450890\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450890\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450890\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable 1\u003C\/span\u003E \u003Cp id=\u0022p-76\u0022 class=\u0022first-child\u0022\u003E\u003Cstrong\u003EParameter values for experimental and optimization setup. The parameters\u003C\/strong\u003E \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003E\u003Cstrong\u003Eother activities\u003C\/strong\u003E\u003C\/sup\u003E, \u003Cem\u003Er, f\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E \u003Cstrong\u003Eand\u003C\/strong\u003E \u003Cem\u003E\u03b8\u003C\/em\u003E \u003Cstrong\u003Erelated to our economic model and are defined in Appendix EC.2\u003C\/strong\u003E\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-77\u0022\u003ETo quantify the benefits of targeting, we consider several ROLD policies that differ in the level of targeting allowed, which we compare over a wide range of values for \u003Cem\u003E\u03c7\u003C\/em\u003E, from 0 to 990\u00d7 the annual GDP per capita in France.\u003Csup\u003E\u003Ca id=\u0022xref-fn-10-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-10\u0022\u003E5\u003C\/a\u003E\u003C\/sup\u003E For each \u003Cem\u003E\u03c7\u003C\/em\u003E value, we calculate all the ROLD policies of interest, and we record separately the economic losses and the number of deaths generated by each policy. The four versions of ROLD we consider are no targeting whatsoever (NO-TARGET), targeting age groups only (AGE), targeting activities only (ACT), or targeting both (AGE-ACT, or simply ROLD when no confusion can arise). To obtain each of the four variants (NO-TARGET, AGE, ACT, AGE-ACT), we run constrained versions of the optimization problem, imposing the following additional constraints:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-19\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-21.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-21.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-20\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-22.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-22.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-21\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-23.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-23.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nWe impose no additional constraints for AGE-ACT. All four variants are allowed to change the confinement policy through time. For each variant, ROLD gets initialized at the solution of a gradient descent algorithm subject to the corresponding constraints.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-26\u0022\u003E\u003Ch2 class=\u0022\u0022\u003E6. Results\u003C\/h2\u003E\u003Cp id=\u0022p-78\u0022\u003EIn this section, we apply our methodology to a case study calibrated with COVID-19 data from the \u00cele-de-France region as described in the previous section. We use the ensuing model to answer our main research questions regarding the efficacy of optimized targeting.\u003C\/p\u003E\u003Cdiv id=\u0022sec-27\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E6.1. How large are the gains from dual targeting?\u003C\/h3\u003E\u003Cp id=\u0022p-79\u0022\u003ETo isolate the benefits of each type of targeting, we compare the four versions of ROLD that differ in the level of targeting allowed, as described in \u003Ca id=\u0022xref-sec-25-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-25\u0022\u003ESection 5.2\u003C\/a\u003E. \u003Ca id=\u0022xref-fig-2-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F2\u0022\u003EFigure 2a\u003C\/a\u003E records each policy\u2019s performance in several problem instances parameterized by the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E. A striking feature is that each of the targeted policies actually \u003Cem\u003EPareto-dominates\u003C\/em\u003E the NO-TARGET policy, and the improvements are significant: relative to NO-TARGET and for same number of deaths, economic losses are reduced by EUR 0-2.9B (0%-35.9%) in AGE, by EUR 0.4B-2.1B (4.5%-49.8%) in ACT, and by EUR 3.3B-5.3B (35.7%-80.0%) in AGE-ACT. This Pareto-dominance is unexpected since it is not explicitly required in our optimization procedure, and it underlines that any form of targeting can lead to significant improvements in terms of \u003Cem\u003Eboth\u003C\/em\u003E health and economic outcomes.\u003C\/p\u003E\u003Cdiv id=\u0022F2\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F2.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022The total number of deaths and the economic losses generated by ROLD policies with different levels of targeting and by the benchmark policies. Panel (a) compares the four versions of ROLD that differ in the level of targeting allowed. Panel (b) compares the ROLD policy that targets age groups and activities with the benchmark policies. Each marker corresponds to a different problem instance parametrized by the cost of death \u0026#x3C7;. We include 128 distinct values of \u0026#x3C7; from 0 to 990\u0026#xD7;, and panel (b) also includes a very large value (\u0026#x3C7; = 1016\u0026#xD7;).\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;The total number of deaths and the economic losses generated by ROLD policies with different levels of targeting and by the benchmark policies. Panel (a) compares the four versions of ROLD that differ in the level of targeting allowed. Panel (b) compares the ROLD policy that targets age groups and activities with the benchmark policies. Each marker corresponds to a different problem instance parametrized by the cost of death \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt;. We include 128 distinct values of \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt; from 0 to 990\u0026lt;em\u0026gt;\u0026#xD7;\u0026lt;\/em\u0026gt;, and panel (b) also includes a very large value (\u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt; = 10\u0026lt;sup\u0026gt;16\u0026lt;\/sup\u0026gt;\u0026lt;em\u0026gt;\u0026#xD7;\u0026lt;\/em\u0026gt;).\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure 2\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F2.medium.gif\u0022 width=\u0022440\u0022 height=\u0022219\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure 2\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F2.medium.gif\u0022 width=\u0022440\u0022 height=\u0022219\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F2.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure 2\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F2.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure 2\u003C\/span\u003E \u003Cp id=\u0022p-80\u0022 class=\u0022first-child\u0022\u003EThe total number of deaths and the economic losses generated by ROLD policies with different levels of targeting and by the benchmark policies. Panel (a) compares the four versions of ROLD that differ in the level of targeting allowed. Panel (b) compares the ROLD policy that targets age groups and activities with the benchmark policies. Each marker corresponds to a different problem instance parametrized by the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E. We include 128 distinct values of \u003Cem\u003E\u03c7\u003C\/em\u003E from 0 to 990\u003Cem\u003E\u00d7\u003C\/em\u003E, and panel (b) also includes a very large value (\u003Cem\u003E\u03c7\u003C\/em\u003E = 10\u003Csup\u003E16\u003C\/sup\u003E\u003Cem\u003E\u00d7\u003C\/em\u003E).\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-81\u0022\u003EWhen comparing the different types of targeting, neither AGE nor ACT Pareto-dominate each other, and neither policy dominates in terms of the total loss objective (\u003Ca id=\u0022xref-fig-3-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F3\u0022\u003EFigure 3a\u003C\/a\u003E). In contrast and crucially, AGE-ACT Pareto-dominates \u003Cem\u003Eall\u003C\/em\u003E other policies, which implies its dominance in terms of the total loss objective. Moreover, targeting both age groups and activities leads to super-additive improvements in almost all cases: for the same number of deaths, AGE-ACT reduces economic losses by more than AGE and ACT \u003Cem\u003Eadded together\u003C\/em\u003E (\u003Ca id=\u0022xref-fig-4-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F4\u0022\u003EFigure 4\u003C\/a\u003E). This suggests that substantial complementarities may be unlocked through targeting \u003Cem\u003Eboth\u003C\/em\u003E age groups and activities, which may not be available under less granular targeting. These results are robust under more problem instances (Appendix Figure EC.2).\u003C\/p\u003E\u003Cdiv id=\u0022F3\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F3.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Total losses generated by ROLD policies with different levels of targeting and by the benchmark policies, at different values of the cost of death \u0026#x3C7;. Panel (a) compares the four versions of ROLD that differ in the level of targeting allowed. Panel (b) compares the ROLD policy that targets age groups and activities with the benchmark policies. Each marker corresponds to a different problem instance parametrized by the cost of death \u0026#x3C7;. We include 128 distinct values of \u0026#x3C7; from 0 to 990\u0026#xD7;\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;Total losses generated by ROLD policies with different levels of targeting and by the benchmark policies, at different values of the cost of death \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt;. Panel (a) compares the four versions of ROLD that differ in the level of targeting allowed. Panel (b) compares the ROLD policy that targets age groups and activities with the benchmark policies. Each marker corresponds to a different problem instance parametrized by the cost of death \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt;. We include 128 distinct values of \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt; from 0 to 990\u0026lt;em\u0026gt;\u0026#xD7;\u0026lt;\/em\u0026gt;\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure 3\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F3.medium.gif\u0022 width=\u0022440\u0022 height=\u0022214\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure 3\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F3.medium.gif\u0022 width=\u0022440\u0022 height=\u0022214\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F3.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure 3\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F3.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure 3\u003C\/span\u003E \u003Cp id=\u0022p-82\u0022 class=\u0022first-child\u0022\u003ETotal losses generated by ROLD policies with different levels of targeting and by the benchmark policies, at different values of the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E. Panel (a) compares the four versions of ROLD that differ in the level of targeting allowed. Panel (b) compares the ROLD policy that targets age groups and activities with the benchmark policies. Each marker corresponds to a different problem instance parametrized by the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E. We include 128 distinct values of \u003Cem\u003E\u03c7\u003C\/em\u003E from 0 to 990\u003Cem\u003E\u00d7\u003C\/em\u003E\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv id=\u0022F4\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F4.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022The super-additivity of ROLD AGE-ACT. The figures compare the improvement from AGE-ACT with the sum of the improvements from AGE and ACT. All improvements are with respect to NO-TARGET. Panel (a) compares the improvements in economic losses for the same number of deaths. Panel (b) compares the improvements in total losses for the same cost of death \u0026#x3C7;.\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;The super-additivity of ROLD AGE-ACT. The figures compare the improvement from AGE-ACT with the sum of the improvements from AGE and ACT. All improvements are with respect to NO-TARGET. Panel (a) compares the improvements in economic losses for the same number of deaths. Panel (b) compares the improvements in total losses for the same cost of death \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt;.\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure 4\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F4.medium.gif\u0022 width=\u0022440\u0022 height=\u0022182\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure 4\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F4.medium.gif\u0022 width=\u0022440\u0022 height=\u0022182\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F4.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure 4\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F4.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure 4\u003C\/span\u003E \u003Cp id=\u0022p-83\u0022 class=\u0022first-child\u0022\u003EThe super-additivity of ROLD AGE-ACT. The figures compare the improvement from AGE-ACT with the sum of the improvements from AGE and ACT. All improvements are with respect to NO-TARGET. Panel (a) compares the improvements in economic losses for the same number of deaths. Panel (b) compares the improvements in total losses for the same cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E.\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-84\u0022\u003ETo confirm the significance of these gains, we also compare ROLD AGE-ACT with various practical benchmark policies in \u003Ca id=\u0022xref-fig-2-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F2\u0022\u003EFigures 2b\u003C\/a\u003E, \u003Ca id=\u0022xref-fig-3-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F3\u0022\u003E3b\u003C\/a\u003E. Benchmarks ICU-t and Hybrid-t AND\/Hybrid-t OR mimic implementations in the U.S. Austin area (\u003Ca id=\u0022xref-ref-17-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-17\u0022\u003EDuque et al. 2020\u003C\/a\u003E) and, respectively, France (\u003Ca id=\u0022xref-ref-32-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-32\u0022\u003ELehot and Borgne 2020\u003C\/a\u003E). These policies switch between a stricter and a relaxed confinement level based on conditions related to hospital admissions and occupancy and the rate of new infections (Appendix EC.5). We also consider two extreme benchmarks corresponding to enforcing \u201cfull confinement\u201d (FC) or remaining \u201cfully open\u201d (FO); these can be expected to perform well when completely prioritizing one of the two metrics of interest, with FC minimizing the number of deaths and FO ensuring low economic losses.\u003C\/p\u003E\u003Cp id=\u0022p-85\u0022\u003EROLD Pareto-dominates all these benchmarks, decreasing economic losses by EUR 5.3B-16.9B (71.0%-82.6%) relative to Hybrid-t AND, by EUR 7.1B-11.6B (62.2%-82.8%) relative to Hybrid-t OR, and by EUR 5.4B-11.6B (62.2%-78.0%) relative to ICU-t for the same number of deaths. Additionally, ROLD meets or exceeds the performance of the two extreme policies: for a sufficiently large \u003Cem\u003E\u03c7\u003C\/em\u003E, ROLD exactly recovers the FC policy, resulting in 890 deaths and economic losses of EUR 27.6B; for a sufficiently low \u003Cem\u003E\u03c7\u003C\/em\u003E, ROLD actually Pareto-dominates the FO policy, reducing the number of deaths by 16,688 (76.7%) and reducing economic losses by EUR 1.6B (65.3%). The latter result, which may seem surprising, is driven by the natural premise captured in our model that deaths and illness generate economic loss because of lost productivity; thus, a smart sequence of confinement decisions can actually improve the economic loss relative to FO. Among all the policies we tested, ROLD AGE-ACT was \u003Cem\u003Ethe only one\u003C\/em\u003E capable of Pareto-dominating the FO benchmark, confirming that dual targeting is critical and powerful.\u003C\/p\u003E\u003Cp id=\u0022p-86\u0022\u003EAnother possible benefit of finer targeting is that it could reduce time in confinement for all groups, even though this is not an explicit objective of ROLD. To check this, we calculate the fraction of time spent by each age group in confinement under each ROLD policy, averaged over the activities relevant to that age group (Appendix EC.6.2). The results are visualized in \u003Ca id=\u0022xref-fig-5-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F5\u0022\u003EFigure 5\u003C\/a\u003E, which depicts boxplots for the fractions of time in confinement across all problem instances parameterized by \u003Cem\u003E\u03c7\u003C\/em\u003E.\u003C\/p\u003E\u003Cdiv id=\u0022F5\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F5.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Average time in confinement for the ROLD policies with different targeting types. Each boxplot depicts the fraction of time the age group spends in confinement under the respective policy averaged over the activities relevant to that age-group, for different problem instances parameterized by \u0026#x3C7;.\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;Average time in confinement for the ROLD policies with different targeting types. Each boxplot depicts the fraction of time the age group spends in confinement under the respective policy averaged over the activities relevant to that age-group, for different problem instances parameterized by \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt;.\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure 5\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F5.medium.gif\u0022 width=\u0022440\u0022 height=\u0022175\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure 5\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F5.medium.gif\u0022 width=\u0022440\u0022 height=\u0022175\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F5.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure 5\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F5.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure 5\u003C\/span\u003E \u003Cp id=\u0022p-87\u0022 class=\u0022first-child\u0022\u003EAverage time in confinement for the ROLD policies with different targeting types. Each boxplot depicts the fraction of time the age group spends in confinement under the respective policy averaged over the activities relevant to that age-group, for different problem instances parameterized by \u003Cem\u003E\u03c7\u003C\/em\u003E.\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-88\u0022\u003EThe dual-targeted AGE-ACT policy is able to reduce the confinement time quite systematically for \u003Cem\u003Eevery\u003C\/em\u003E age group, relative to \u003Cem\u003Eall\u003C\/em\u003E other policies. Specifically, it results in the lowest confinement time for every age group in 70% of all problem instances when compared with NO-TARGET, in 60% of instances when compared with AGE, in 83% of instances when compared with ACT, and in 50% of instances when compared with \u003Cem\u003Eall\u003C\/em\u003E other policies. Moreover, the fraction of confinement time achieved by AGE-ACT is within 5% (in absolute terms) from the lowest confinement time achieved by any policy for every age group, in 76% of all instances; within 10% in 80% of the instances; and within 14% in all instances. Thus, even when the dual-targeted policy confines certain age groups more, it does not do so by much. These outcomes are quite unexpected as they are not something that the ROLD framework explicitly optimizes for, but rather a \u003Cem\u003Eby-product\u003C\/em\u003E of a dual-targeted confinement policy that minimizes the total loss objective (11).\u003Csup\u003E\u003Ca id=\u0022xref-fn-11-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-11\u0022\u003E6\u003C\/a\u003E\u003C\/sup\u003E\u003C\/p\u003E\u003Cp id=\u0022p-89\u0022\u003EIt is worth noting that although ROLD AGE-ACT reduces confinements for every age group compared to less targeted policies, it does not do so uniformly. Instead, it can lead to a larger discrepancy in the confinements faced by different age groups: those aged 10-59 are generally more confined than those aged 0-9 or 60+.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-28\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E6.2. How do gains arise from dual targeting?\u003C\/h3\u003E\u003Cp id=\u0022p-90\u0022\u003EWe examine the structure of the optimal ROLD AGE-ACT confinement decisions. \u003Ca id=\u0022xref-fig-6-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F6\u0022\u003EFigure 6\u003C\/a\u003E visualizes the optimized confinement policy for the value \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u00d7, which is in the mid-range of estimates used in the economics literature on COVID-19 (\u003Ca id=\u0022xref-ref-4-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-4\u0022\u003EAlvarez et al. 2020\u003C\/a\u003E) and is representative of the behavior we observe across all experiments.\u003C\/p\u003E\u003Cdiv id=\u0022F6\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F6.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022The optimized ROLD AGE-ACT policies for a problem instance with a 90-day optimization horizon starting on October 21, 2020, and with cost of death \u0026#x3C7; = 50\u0026#xD7;. (See Appendix EC.6.4 for optimized ROLD policies with a 360-day optimization horizon.) From top to bottom, the seven panels depict the time evolution for the occupation of hospital and ICU beds, the number of actively infectious individuals and the cumulative number of deceased individuals in the population, and the confinement policy imposed by ROLD in each age group and activity. In panels 3-7, the values correspond to the activity levels allowed and are color-coded so that darker shades capture a stricter confinement.\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;The optimized ROLD AGE-ACT policies for a problem instance with a 90-day optimization horizon starting on October 21, 2020, and with cost of death \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt; = 50\u0026lt;em\u0026gt;\u0026#xD7;\u0026lt;\/em\u0026gt;. (See Appendix EC.6.4 for optimized ROLD policies with a 360-day optimization horizon.) From top to bottom, the seven panels depict the time evolution for the occupation of hospital and ICU beds, the number of actively infectious individuals and the cumulative number of deceased individuals in the population, and the confinement policy imposed by ROLD in each age group and activity. In panels 3-7, the values correspond to the activity levels allowed and are color-coded so that darker shades capture a stricter confinement.\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure 6\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F6.medium.gif\u0022 width=\u0022386\u0022 height=\u0022440\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure 6\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F6.medium.gif\u0022 width=\u0022386\u0022 height=\u0022440\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F6.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure 6\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F6.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure 6\u003C\/span\u003E \u003Cp id=\u0022p-91\u0022 class=\u0022first-child\u0022\u003EThe optimized ROLD AGE-ACT policies for a problem instance with a 90-day optimization horizon starting on October 21, 2020, and with cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u003Cem\u003E\u00d7\u003C\/em\u003E. (See Appendix EC.6.4 for optimized ROLD policies with a 360-day optimization horizon.) From top to bottom, the seven panels depict the time evolution for the occupation of hospital and ICU beds, the number of actively infectious individuals and the cumulative number of deceased individuals in the population, and the confinement policy imposed by ROLD in each age group and activity. In panels 3-7, the values correspond to the activity levels allowed and are color-coded so that darker shades capture a stricter confinement.\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-92\u0022\u003EGenerally, the ROLD policy maintains high activity levels for those groups with a high ratio of marginal economic value to total social contacts in the activity, i.e., a high\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-22\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-29.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-29.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E(\u003Cem\u003Et\u003C\/em\u003E)) is the economic value created by an individual in age group \u003Cem\u003Eg\u003C\/em\u003E when activity levels are \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E(\u003Cem\u003Et\u003C\/em\u003E). For example, in \u003Ckbd\u003Ework\u003C\/kbd\u003E, ROLD completely opens up the 40-69 y.o. groups, while confining the 20-39 y.o. groups during the first two weeks and the 10-19 y.o. group for the first ten weeks. This is explainable since the 40-69 y.o. groups produce the highest econ-to-contacts-ratio in \u003Ckbd\u003Ework\u003C\/kbd\u003E, while younger groups have progressively lower ratios. Similarly, ROLD prioritizes activity in \u003Ckbd\u003Etransport\u003C\/kbd\u003E, then \u003Ckbd\u003Eother\u003C\/kbd\u003E, then \u003Ckbd\u003Eleisure\u003C\/kbd\u003E, in accordance with the relative econ-to-contacts-ratio of these activities.\u003C\/p\u003E\u003Cp id=\u0022p-93\u0022\u003ETo understand how \u003Cem\u003Ecomplementarities\u003C\/em\u003E arise in this context, note that the ability to separately target age groups and activities allows the ROLD policy to fully exploit the fact that distinct age groups may be responsible for the largest econ-to-contacts-ratio in different activities. As an example, the 20-69 y.o. groups have the highest ratio in \u003Ckbd\u003Ework\u003C\/kbd\u003E, whereas the 0-19 y.o. and 70+ y.o. groups have the highest ratio in \u003Ckbd\u003Eleisure\u003C\/kbd\u003E. Accordingly, we see that ROLD coordinates confinements to account for this: groups 20-69 y.o. remain more open in \u003Ckbd\u003Ework\u003C\/kbd\u003E but face confinement in \u003Ckbd\u003Eleisure\u003C\/kbd\u003E for up to the first ten weeks; on the contrary, the 10-19 y.o. group is confined in \u003Ckbd\u003Ework\u003C\/kbd\u003E for a long period while remaining open in \u003Ckbd\u003Eleisure\u003C\/kbd\u003E, and the 70+ y.o. groups also remain open in \u003Ckbd\u003Eleisure\u003C\/kbd\u003E. These complementary confinement schedules allow ROLD to reduce both the number of deaths and economic losses, with the important added benefit that no age group is completely confined.\u003C\/p\u003E\u003Cp id=\u0022p-94\u0022\u003ETo generalize the above insights and gain a better understanding of the ROLD policy, we also take a different approach and train interpretable machine learning models (regression trees) to predict the optimal ROLD confinement decisions as a function of several features that depend on salient epidemiological and economic parameters.\u003C\/p\u003E\u003Cp id=\u0022p-95\u0022\u003EMore specifically, we generate 26,140 problem instances, with (i) randomly drawn values for the transmission rate \u003Cem\u003E\u03b2\u003C\/em\u003E, the probabilities of an infectious individual having severe symptoms {\u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Ess,g\u003C\/em\u003E\u003C\/sub\u003E}\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E and the parameter of the social mixing model \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E = \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E2\u003C\/sub\u003E = \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E1,2\u003C\/sub\u003E, and (ii) a finite set of values for the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E, the ICU capacity, and the sensitivity of economic value on confinement in \u003Ckbd\u003Etransport\u003C\/kbd\u003E, \u003Ckbd\u003Eleisure\u003C\/kbd\u003E and \u003Ckbd\u003Eother\u003C\/kbd\u003E, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E. We solve these instances using ROLD and then we fit a decision tree to predict the ROLD-optimized activity levels \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-39\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-39.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-39.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E based on 22 features. We include as features several important problem parameters (e.g., \u003Cem\u003E\u03b2, \u03b1\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E = \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E2\u003C\/sub\u003E = \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E1,2\u003C\/sub\u003E, ICU capacity, economic parameters) as well as derived features based on parameters and SEIR state values (e.g., econ-to-contacts-ratio, ICU utilization and admission rate, infection rates, etc.). Some of these features are targeted, meaning they take different values at a given time step for different age groups or activities (e.g., econ-to-contacts-ratio), whereas others are non-targeted, such as the transmission rate \u003Cem\u003E\u03b2\u003C\/em\u003E, the time \u003Cem\u003Et\u003C\/em\u003E or the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E. The details of the data generation and fitting procedures can be found in Appendix EC.6.3, and the full set of features is in \u003Ca id=\u0022xref-table-wrap-9-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T9\u0022\u003ETables EC.8\u003C\/a\u003E and \u003Ca id=\u0022xref-table-wrap-10-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T10\u0022\u003EEC.9\u003C\/a\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-96\u0022\u003E\u003Ca id=\u0022xref-fig-7-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F7\u0022\u003EFigure 7\u003C\/a\u003E displays the depth-four tree obtained by fitting using the entire feature set, together with the resulting root-mean-squared-error (RMSE). This simple tree predicts the optimal ROLD activity levels quite well (RMSE of 0.15) and it confirms our core insight that the econ-to-contacts-ratio is the most salient feature when targeting confinements, as it is used as a split variable in the root node of the tree and also subsequently used for splits in two sub-trees, with higher ratios leading to higher activity levels. The tree also confirms that time and the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E are relevant: the optimal ROLD policy tends to enforce stricter confinements in earlier periods and subsequently relaxes these through time, and the confinements become stricter for higher values of \u003Cem\u003E\u03c7\u003C\/em\u003E. The only other feature appearing in the tree is R-perc, which quantifies the total number of individuals in a recovered state, as a fraction of the overall population; this is a natural measure of the level of herd immunity, and the ROLD policy generally seems aligned with it in an intuitive way, enforcing stricter confinements at lower levels of herd immunity.\u003C\/p\u003E\u003Cdiv id=\u0022F7\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F7.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Decision tree of depth four approximating the optimized ROLD confinement decisions trained on a total of 26,140 problem instances with a horizon of T = 90 days (a total of 8,232,210 samples). The nodes are color-coded based on the activity level, with darker colors corresponding to stricter confinement.\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;Decision tree of depth four approximating the optimized ROLD confinement decisions trained on a total of 26,140 problem instances with a horizon of \u0026lt;em\u0026gt;T\u0026lt;\/em\u0026gt; = 90 days (a total of 8,232,210 samples). The nodes are color-coded based on the activity level, with darker colors corresponding to stricter confinement.\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure 7\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F7.medium.gif\u0022 width=\u0022440\u0022 height=\u0022348\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure 7\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F7.medium.gif\u0022 width=\u0022440\u0022 height=\u0022348\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F7.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure 7\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F7.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure 7\u003C\/span\u003E \u003Cp id=\u0022p-97\u0022 class=\u0022first-child\u0022\u003EDecision tree of depth four approximating the optimized ROLD confinement decisions trained on a total of 26,140 problem instances with a horizon of \u003Cem\u003ET\u003C\/em\u003E = 90 days (a total of 8,232,210 samples). The nodes are color-coded based on the activity level, with darker colors corresponding to stricter confinement.\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-98\u0022\u003EWe also quantify the importance of the various features by calculating their permutation importance scores, a commonly used metric that measures importance as the increase in model prediction error after the values of the respective feature are randomly permuted. For the tree in \u003Ca id=\u0022xref-fig-7-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F7\u0022\u003EFigure 7\u003C\/a\u003E, the scores for econ-to-contacts-ratio, time, R-perc, and \u003Cem\u003E\u03c7\u003C\/em\u003E are respectively 0.436, 0.019, 0.004, 0.002, and all the other features have scores of zero. The econ-to-contacts-ratio feature thus has by far the largest permutation importance score of all features. \u003Ca id=\u0022xref-fig-18-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F18\u0022\u003EFigure EC.4\u003C\/a\u003E in the Appendix confirms similar results for a tree of depth 10, and \u003Ca id=\u0022xref-fig-19-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F19\u0022\u003EFigure EC.5\u003C\/a\u003E documents an additional exercise that further shows econ-to-contacts-ratio as the most salient of all targeted features.\u003C\/p\u003E\u003Cp id=\u0022p-99\u0022\u003EThe importance of econ-to-contacts-ratio provides increased transparency into how targeted confinement decisions could be made. A policy maker can explain optimized confinement decisions based on characteristics relating to economic value and public health risk, which is much less contentious than justifying confinement policy based merely on age or activity \u2013 to that point, note that the depth-four tree does not split on either the age or the activity.\u003C\/p\u003E\u003Cp id=\u0022p-100\u0022\u003EThe Appendix contains additional results to complement this section, including a theoretical justification for the salience of the econ-to-contacts-ratio derived in a simplified model (Appendix EC.7), a more detailed discussion of the optimal ROLD confinement policy (Appendix EC.6.4), and a quantification of the value of targeted testing (Appendix EC.6.5).\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-29\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003E6.3. The impact of limited disparity requirements\u003C\/h3\u003E\u003Cp id=\u0022p-101\u0022\u003EThat targeted policies confine some age groups more than others could be perceived as unfair treatment, so it is important to quantify how an intervention\u2019s effectiveness is impacted when requiring less differentiation across age groups. To capture fairness considerations, we embed a set of \u201climited disparity\u201d constraints in ROLD that allow the activity levels of distinct age groups to differ by at most \u0394 in absolute terms, in each activity and at any time (see definition in Appendix EC.6.6). \u0394 = 0 corresponds to a strictly non-discriminatory policy, whereas a larger \u0394 allows more targeting, and \u0394 = 1 captures a fully targeted policy. For every \u0394, we record the total loss incurred by a ROLD policy with the limited disparity constraints and the increase in total loss relative to a fully targeted ROLD policy. We repeat the experiment for different problem instances parametrized by \u003Cem\u003E\u03c7\u003C\/em\u003E, and \u003Ca id=\u0022xref-fig-8-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F8\u0022\u003EFigure 8\u003C\/a\u003E depicts boxplots of all the relative increases in total loss, as a function of \u0394.\u003C\/p\u003E\u003Cdiv id=\u0022F8\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F8.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022The impact of limited disparity requirements. The plot shows the relative (%) increase in total loss generated by a ROLD policy, compared to a fully targeted policy, as a function of the disparity parameter \u0026#x394; that measures the maximum allowed difference in activity levels for distinct age groups. The experiments are run using several values of \u0026#x3C7;, which are used to generate each of the boxplots.\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;The impact of limited disparity requirements. The plot shows the relative (%) increase in total loss generated by a ROLD policy, compared to a fully targeted policy, as a function of the disparity parameter \u0026#x394; that measures the maximum allowed difference in activity levels for distinct age groups. The experiments are run using several values of \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt;, which are used to generate each of the boxplots.\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure 8\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F8.medium.gif\u0022 width=\u0022440\u0022 height=\u0022298\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure 8\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F8.medium.gif\u0022 width=\u0022440\u0022 height=\u0022298\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F8.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure 8\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F8.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure 8\u003C\/span\u003E \u003Cp id=\u0022p-102\u0022 class=\u0022first-child\u0022\u003EThe impact of limited disparity requirements. The plot shows the relative (%) increase in total loss generated by a ROLD policy, compared to a fully targeted policy, as a function of the disparity parameter \u0394 that measures the maximum allowed difference in activity levels for distinct age groups. The experiments are run using several values of \u003Cem\u003E\u03c7\u003C\/em\u003E, which are used to generate each of the boxplots.\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-103\u0022\u003EThe results suggest that limited disparity requirements may be costly: on average, completely eliminating disparity in confinements would increase the total losses by EUR 1.2B (21.6%) and produce an additional 506 deaths (16.6%) and an extra EUR 0.5B of economic losses (18.9%) compared to a fully targeted policy. In certain instances, the increase in total loss could be as high as 63%. The high losses persist even when some limited discrepancy is allowed, dropping at an initially slow rate as \u0394 increases from 0 and eventually at a slightly faster rate as it approaches 1. This suggests that to fully leverage the benefits of targeting, a high level of disparity must be accepted, but reasonable trade-offs can be achieved with some intermediate disparity.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-30\u0022\u003E\u003Ch2 class=\u0022\u0022\u003E7. Discussion\u003C\/h2\u003E\u003Cp id=\u0022p-104\u0022\u003EOur case study suggests that an optimized intervention targeting both age groups \u003Cem\u003Eand\u003C\/em\u003E activities carries significant promise for alleviating a pandemic\u2019s health, economic and even psychological burden, but also points to certain challenges that require care in a real-world setting.\u003C\/p\u003E\u003Cdiv id=\u0022sec-31\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003EWhy consider optimized dual-targeted interventions?\u003C\/h3\u003E\u003Cp id=\u0022p-105\u0022\u003EThe first reason are the significantly better health \u003Cem\u003Eand\u003C\/em\u003E economic outcomes: for the same or a lower number of deaths, dual-targeted confinements can reduce economic losses more than any of the simpler interventions that uniformly confine age groups or activities. Furthermore, the super-additive gains imply that significant synergies can be generated through finer targeting, with the ability to target along activities improving the effectiveness of targeting along age groups, and vice-versa. Dual-targeting also may enable all age groups to remain more active, resembling normal life more closely compared to less differentiated confinements. This could result in more socially acceptable restrictions, and a more appealing policy intervention overall.\u003C\/p\u003E\u003Cp id=\u0022p-106\u0022\u003EThe second reason is the interpretable nature of the optimized targeted confinement policy, which is consistent with a simple \u201cbang-for-the-buck\u201d rule: impose less confinement on group-activity pairs that generate a relatively high economic value prorated by (activity-specific) social contacts. This simple intuition combined with the reliance on just a few activity levels are appealing practical features, as they provide transparency into how targeted confinement decisions could be made.\u003C\/p\u003E\u003Cp id=\u0022p-107\u0022\u003ELastly, we note that although dual targeting allows and can result in differences in confinements across age groups, such interventions are actually not far from many real-world policy implementations, which have been more or less explicit in their age-based discrimination. Dual targeting can arise \u003Cem\u003Eimplicitly\u003C\/em\u003E in interventions that only seem to target activities. As an example, France implemented a population-wide 6 p.m. to 6 a.m. curfew during the first few months of 2021 (\u003Ca id=\u0022xref-ref-42-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-42\u0022\u003EReuters 2021\u003C\/a\u003E), while maintaining school and work activities largely de-confined. This is effectively implementing restrictions similar to ROLD AGE-ACT: since a typical member of the 20-65 y.o. group is engaged in \u003Ckbd\u003Ework\u003C\/kbd\u003E until the start of the curfew, their \u003Ckbd\u003Eleisure\u003C\/kbd\u003E and \u003Ckbd\u003Eother\u003C\/kbd\u003E activities are implicitly limited; moreover, since most individuals aged above 65 are not in active employment, they are not that restricted in these last two activities.\u003C\/p\u003E\u003Cp id=\u0022p-108\u0022\u003EThese examples show that some amount of targeting of activities \u003Cem\u003Eand\u003C\/em\u003E age groups is already in place and is perhaps unavoidable for effective pandemic management. Given this state of affairs, our framework highlights the significant benefits in explicitly and transparently modelling targeting and identifying the interventions that rigorously optimize overall societal welfare, given some allowable amount of differentiation across age groups.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-32\u0022 class=\u0022subsection\u0022\u003E\u003Ch3\u003EChallenges and limitations\u003C\/h3\u003E\u003Cp id=\u0022p-109\u0022\u003EAn immediate practical challenge is data availability. Social contact matrices by age group and activity may be available from surveys on social behavior, which have been conducted in a number of countries; however, further data collection might be required to obtain these matrices for more refined population group or activity definitions. Similarly, economic data is reported by industry activities, but we are not aware of a dataset that splits economic value into separate (group, activity) contributions. Disparate data sources may be difficult to align: for example, social contact surveys and Google mobility reports use different activity categories, which requires non-trivial fitting (Appendix EC.4).\u003C\/p\u003E\u003Cp id=\u0022p-110\u0022\u003EAvailability of data also constrains our model\u2019s structure in several ways. Social contacts between age groups only depend on confinements in the same activity, since the available social contacts dataset (\u003Ca id=\u0022xref-ref-8-4\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-8\u0022\u003EB\u00e9raud et al. 2015\u003C\/a\u003E) only reports interactions in the same activity. However, contacts occur as individuals are engaged in different activities (e.g., a services industry professional interacts during work with individuals who are engaged in leisure activities). A more refined contact mixing model that captures such interactions would be more appropriate for this study, provided that relevant data are available. Our economic model similarly ignores cross effects, such as young age groups engaged in school producing value in conjunction with educational staff engaged in work.\u003C\/p\u003E\u003Cp id=\u0022p-111\u0022\u003EAnother challenge with targeted interventions is the perception that they lead to unfair outcomes, as certain age groups face more confinement than others. Such discrimination does arise in the optimized dual-targeted policies; our framework partially addresses these concerns through explicit constraints that limit the disparities across groups. Our requirement that limited disparity hold for every time period and every activity is quite strict, and a looser requirement based, e.g., on time-average confinements could lead to smaller incremental losses. Alternatively, one can impose fairness requirements based on the intervention\u2019s \u003Cem\u003Eoutcomes\u003C\/em\u003E, e.g., requiring that the health or economic losses faced by different groups satisfy certain axiomatic fairness properties \u003Ca id=\u0022xref-ref-50-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-50\u0022\u003EYoung (1994)\u003C\/a\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-112\u0022\u003EAlthough we focus on confinement policies, a direction for future research is to investigate how these can be optimally combined with other types of targeted interventions. Appendix EC.6 reports experiments where we optimize a targeted policy based on confinements and randomized testing and quarantining. The framework is sufficiently flexible to accommodate interventions such as contact tracing and also vaccinations, although a careful implementation would require work beyond the scope of this article.\u003C\/p\u003E\u003Cp id=\u0022p-113\u0022\u003EFrom a more technical standpoint, future research could be devoted to directly learning an optimal intervention from epidemiological and economic outcomes, bypassing the estimation-optimization procedure we adopted here. Work could also be devoted to deriving a theoretical justification for the importance of the econ-to-contacts-ratio by generalizing our simple analysis in Appendix EC.7. Lastly, future work could focus on deriving tractable upper bounds for the performance of controlled SEIR models that could be used to benchmark various interventions applied in practice or proposed in the academic literature.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section data-availability\u0022 id=\u0022sec-33\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EData Availability\u003C\/h2\u003E\u003Cp id=\u0022p-114\u0022\u003EAll data referred to in the manuscript are public and URL references to the datasets are included in the manuscript.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv id=\u0022sec-34\u0022 class=\u0022subsection\u0022\u003E\u003Cp id=\u0022p-115\u0022\u003E\u003Cstrong\u003EThis page is intentionally blank. Proper e-companion title page, with INFORMS branding and exact metadata of the main paper, will be produced by the INFORMS office when the issue is being assembled\u003C\/strong\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-35\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EE-companion\u003C\/h2\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-36\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EEC.1. Dynamics of the Controlled SEIR Epidemic Model\u003C\/h2\u003E\u003Cp id=\u0022p-116\u0022\u003EWe write down a set of discrete time dynamics for the controlled SEIR model. We use notation \u0394\u003Cem\u003EZ\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) to denote \u003Cem\u003EZ\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E + 1) \u2212 \u003Cem\u003EZ\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E). For all groups \u003Cem\u003Eg\u003C\/em\u003E \u2208 \ud835\udca2, we have:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-23\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-32.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-32.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-24\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-33.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-33.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-25\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-34.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-34.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-26\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-35.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-35.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-27\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-36.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-36.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-28\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-37.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-37.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-29\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-38.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-38.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-30\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-39.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-39.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-31\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-40.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-40.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-32\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-41.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-41.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-33\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-42.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-42.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nIn (EC.5) and (EC.8), note that we do not have terms for the population turned away from hospital\/ICU which may eventually recover. Instead, we assume the turned away patients will go into the deceased state. In (EC.11), we are assuming that if a patient is turned away from the ICU, they transition into deceased, instead of being allocated a hospital bed if one is available.\u003C\/p\u003E\u003Cp id=\u0022p-117\u0022\u003EWe now provide justification for how we account for social contacts and, in particular, for the expressions in (EC.2) and (EC.3). We note that individuals in \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-40\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-40.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-40.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E can interact with members of \u003Cem\u003ES\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, \u003Cem\u003EE\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, \u003Cem\u003EI\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E and \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E. Fix a person \u003Cem\u003Ei\u003C\/em\u003E in age group \u003Cem\u003Eg\u003C\/em\u003E \u2208 \ud835\udca2, in state \u003Cem\u003ES\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E. Then:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-34\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-43.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-43.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-35\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-44.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-44.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-36\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-45.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-45.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-37\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-46.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-46.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-38\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-47.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-47.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-39\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-48.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-48.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-40\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-49.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-49.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nIn (EC.16) we use the following reasoning. Having fixed person \u003Cem\u003Ei\u003C\/em\u003E in age group \u003Cem\u003Eg\u003C\/em\u003E, (a) a contact with a randomly chosen individual in group \u003Cem\u003Eh\u003C\/em\u003E will result in person \u003Cem\u003Ei\u003C\/em\u003E getting infected with probability \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-41\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-41.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-41.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, and (b) the number of person \u003Cem\u003Ei\u003C\/em\u003E\u2019s contacts with individuals in age group \u003Cem\u003Eh\u003C\/em\u003E is given by \u003Cem\u003Ec\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg,h\u003C\/em\u003E\u003C\/sub\u003E = \u003Cem\u003Ec\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg,h\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003El\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003El\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eh\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E)). Finally, person \u003Cem\u003Ei\u003C\/em\u003E getting infected as the result of a contact with someone from group \u003Cem\u003Eh\u003C\/em\u003E is considered to be an independent event across different contacts, therefore we raise the probability of no infection from a contact to the power of the number of contacts. (EC.17) and (EC.18) follow from linear approximations.\u003C\/p\u003E\u003Cp id=\u0022p-118\u0022\u003EBy taking the expectation of random variable\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-41\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-50.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-50.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwe retrieve the expressions in (EC.2) and (EC.3).\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-37\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EEC.2. Details of the Economic Model\u003C\/h2\u003E\u003Cp id=\u0022p-119\u0022\u003EAs discussed in \u003Ca id=\u0022xref-sec-17-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-17\u0022\u003ESection 3.4\u003C\/a\u003E, economic losses come from three separate sources:\u003C\/p\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-38\u0022\u003E\u003Ch3\u003EEffect of confinement\u003C\/h3\u003E\u003Cp id=\u0022p-120\u0022\u003ETo account for confinement in the non-quarantined population, we make the economic value generated per day by an individual in group \u003Cem\u003Eg\u003C\/em\u003E in the remaining (non-quarantined) SEIR chambers explicitly depend on the enforced confinement in the population. Recall that for a group \u003Cem\u003Eg\u003C\/em\u003E, the activity levels \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E specify the level of each activity allowed for that group as compared to normal course, and \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E = [\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E]\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E. We denote the economic value generated by a member of \u003Cem\u003Eg\u003C\/em\u003E per day by \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E). We remark that \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E1\u003C\/strong\u003E) corresponds to the economic value generated by an individual under normal circumstances.\u003C\/p\u003E\u003Cp id=\u0022p-121\u0022\u003EThe \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E) specific to a group can be of two types: (a) wages from employment and (b) future wages from employment due to schooling. Naturally, depending on the age group, both, one, or neither of these will actually contribute to economic value. Distinguishing whether the specific group is comprised of school age, employable or retired population, we define\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-42\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-51.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-51.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nWe break down the definitions of \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-42\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-42.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-42.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-43\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-43.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-43.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E below:\u003C\/p\u003E\u003Cul class=\u0022list-unord \u0022 id=\u0022list-3\u0022\u003E\u003Cli id=\u0022list-item-9\u0022\u003E\u003Cp id=\u0022p-122\u0022\u003E\u003Cstrong\u003EValue from employment \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-44\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-44.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-44.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/strong\u003E The value generated from employment is a function of the confinement level in the \u003Ckbd\u003Ework\u003C\/kbd\u003E activity, but also of the confinement levels in \u003Ckbd\u003Eleisure\u003C\/kbd\u003E, \u003Ckbd\u003Etransport\u003C\/kbd\u003E, as well as \u003Ckbd\u003Eother\u003C\/kbd\u003E activities. As an example, we expect the economic value generated by those employed in restaurants, retail stores, etc. to depend on foot traffic levels, which in turn are driven by the confinement levels in \u003Ckbd\u003Eleisure\u003C\/kbd\u003E, \u003Ckbd\u003Etransport\u003C\/kbd\u003E and \u003Ckbd\u003Eother\u003C\/kbd\u003E activities across all groups.\u003C\/p\u003E\u003Cp id=\u0022p-123\u0022\u003EOur model for employment value is a linear parametrization of these confinement decisions; specifically, \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-45\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-45.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-45.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E is linear in \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-46\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-46.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-46.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and the average of \u003Cem\u003El\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Etransport\u003C\/kbd\u003E\u003C\/sup\u003E, \u003Cem\u003El\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Eleisure\u003C\/kbd\u003E\u003C\/sup\u003E and \u003Cem\u003El\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Eother\u003C\/kbd\u003E\u003C\/sup\u003E across these three activities and all groups \u003Cem\u003Eg\u003C\/em\u003E \u2208 \ud835\udca2:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-43\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-52.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-52.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nAdditionally, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Ework\u003C\/kbd\u003E\u003C\/sup\u003E, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E and \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Efixed\u003C\/sup\u003E are activity level sensitivity parameters such that \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Ework\u003C\/kbd\u003E\u003C\/sup\u003E\u00b7 1 + \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E \u00b71 + \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Efixed\u003C\/sup\u003E = 1; under fully open activity, they induce a multiplier of 1 in (EC.20). Then \u003Cem\u003Ew\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E measures the overall daily employment value of a typical member of group \u003Cem\u003Eg\u003C\/em\u003E under no confinement, and is equal to \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-47\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-47.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-47.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-124\u0022\u003EWe estimate the coefficients of this model from data, as we describe in detail in \u003Ca id=\u0022xref-sec-47-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-47\u0022\u003ESection EC.4\u003C\/a\u003E.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-10\u0022\u003E\u003Cp id=\u0022p-125\u0022\u003E\u003Cstrong\u003EValue from schooling \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-48\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-48.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-48.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/strong\u003E A day of schooling for the individuals in relevant groups results in economic value, equal to a day of wages that members of these groups would gain in the future. We use the salary of the 20-29 year-old group multiplied by a factor, and we discount for a number of years corresponding to the difference between the midpoint of the age group and the beginning of the 20-29 year-old group. For instance, for the 0-9 year olds we discount over 15 years, and for 10-19 year olds we discount over five years. The discounting factor we apply is thus\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-44\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-53.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-53.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u003Cem\u003Er\u003C\/em\u003E is the discount rate. We further multiply the wage by \u003Cem\u003Ef\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, which is the fraction of group \u003Cem\u003Eg\u003C\/em\u003E that is in school.\u003Csup\u003E\u003Ca id=\u0022xref-fn-12-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-12\u0022\u003E7\u003C\/a\u003E\u003C\/sup\u003E Lastly, we also use a multiplicative factor \u003Cem\u003E\u03b8\u003C\/em\u003E for sensitivity analysis: \u003Cem\u003E\u03b8\u003C\/em\u003E reflects that an additional day of schooling may have a multiplier effect in future wages, as well as the fact that schooling can be continued online during lockdowns. We provide ranges for \u003Cem\u003E\u03b8\u003C\/em\u003E in \u003Ca id=\u0022xref-sec-25-2\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-25\u0022\u003ESection 5.2\u003C\/a\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-126\u0022\u003EThus, the definition for value of school days is\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-45\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-54.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-54.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003C\/p\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-39\u0022\u003E\u003Ch3\u003EEffect of quarantine, illness and deaths during the pandemic\u003C\/h3\u003E\u003Cp id=\u0022p-127\u0022\u003EWe capture the economic effect of quarantine, hospitalization, and death during the pandemic by assuming that if at some time period an individual in group \u003Cem\u003Eg\u003C\/em\u003E is in one of the SEIR chambers \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-49\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-49.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-49.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, \u003Cem\u003EH\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, \u003Cem\u003EICU\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, \u003Cem\u003ED\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, then they generate no economic value. At the same time, we assume that individuals in \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-50\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-50.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-50.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E generate economic value as they would under no confinement.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-40\u0022\u003E\u003Ch3\u003EEffect of lost future wages due to death\u003C\/h3\u003E\u003Cp id=\u0022p-128\u0022\u003EWe account for a deceased individual\u2019s lost wages which they would have earned from their current age until retirement age, given the prevailing wage curve under normal circumstances, {\u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E1\u003C\/strong\u003E) } \u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E. For group \u003Cem\u003Eg\u003C\/em\u003E, we set the current age to the midpoint of the age group. We discount the resulting cash flows by an annualized interest rate. We denote the resulting lost wages amount by\u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-51\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-51.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-51.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-129\u0022\u003EFor instance, for someone in age group 30-39 y.o., we calculate this cash flow by\u003Csup\u003E\u003Ca id=\u0022xref-fn-13-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-13\u0022\u003E8\u003C\/a\u003E\u003C\/sup\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-46\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-55.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-55.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nLast, we define a quantity \u003Cem\u003EV\u003C\/em\u003E which represents the economic value that would be generated across all groups \u003Cem\u003Eg\u003C\/em\u003E\u2208 \ud835\udca2, during the time of the pandemic, under a \u201cno pandemic\u201d scenario. More precisely, to calculate \u003Cem\u003EV\u003C\/em\u003E we assume that at time \u003Cem\u003Et\u003C\/em\u003E = 0 all the infected population is instantaneously healed and able to generate full economic value \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E1\u003C\/strong\u003E). Thus,\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-47\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-56.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-56.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nNote that this term is a constant and does not depend on the policy followed by the policy maker.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-41\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EEC.3. Algorithmic Details for ROLD\u003C\/h2\u003E\u003Cp id=\u0022p-130\u0022\u003EIn this section, we clarify the algorithmic details of the linearization-optimization procedure described in \u003Ca id=\u0022xref-sec-19-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-19\u0022\u003ESection 4\u003C\/a\u003E. We first focus on how we build a linear model given \u003Cem\u003Ek\u003C\/em\u003E, \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ek\u003C\/em\u003E\u003C\/sub\u003E and \u003Cstrong\u003E\u003Cem\u003E\u00fb\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ek\u003C\/em\u003E:\u003Cem\u003ET\u003C\/em\u003E \u22121\u003C\/sub\u003E.\u003C\/p\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-42\u0022\u003E\u003Ch3\u003EEC.3.1. Linearized Dynamics\u003C\/h3\u003E\u003Cp id=\u0022p-131\u0022\u003EIn Step 2, the algorithm builds an approximation of the state dynamics that is linear in the controls \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ek\u003C\/em\u003E\u003C\/sub\u003E, \u2026, \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003ET\u003C\/em\u003E \u22121\u003C\/sub\u003E. Here, we compute the coefficients for each \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E explicitly. We introduce the notation:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-48\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-57.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-57.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-49\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-58.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-58.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-50\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-59.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-59.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere matrix \u003Cstrong\u003E\u003Cem\u003EA\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E has dimensions |\ud835\udca2||\ud835\udcb3| \u00d7 |\ud835\udca2||\ud835\udcb3|, matrix \u003Cstrong\u003E\u003Cem\u003EB\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E has dimensions |\ud835\udca2||\ud835\udcb3| \u00d7 |\ud835\udca2||\ud835\udcb0|, and vector \u003Cstrong\u003E\u003Cem\u003Ec\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E has dimensions |\ud835\udca2||\ud835\udcb3| \u00d7 1. With this, we have\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-51\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-60.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-60.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nWe can then express the state \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E as\u003Csup\u003E\u003Ca id=\u0022xref-fn-14-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-14\u0022\u003E9\u003C\/a\u003E\u003C\/sup\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-52\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-61.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-61.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nIt is now possible to express both the objective and the constraints linearly in the decisions \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-43\u0022\u003E\u003Ch3\u003EEC.3.2. Constraint Coefficients\u003C\/h3\u003E\u003Cp id=\u0022p-132\u0022\u003EWe can write each of the constraints (5), (6) and (7) in the form\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-53\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-62.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-62.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nfor some (time-invariant) \u003Cstrong\u003E\u003Cem\u003E\u03b3\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ex\u003C\/em\u003E\u003C\/sub\u003E, \u003Cstrong\u003E\u003Cem\u003E\u03b3\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/sub\u003E. Since \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E is linear in \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ek\u003C\/em\u003E\u003C\/sub\u003E, \u2026, \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u22121\u003C\/sub\u003E, to represent one such constraint we just need to store the coefficients corresponding to all decision variables (i.e., \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ek\u003C\/em\u003E\u003C\/sub\u003E, \u2026, \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003ET\u003C\/em\u003E \u22121\u003C\/sub\u003E) and the free terms\/constants that appear in \u003Cem\u003EL\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-133\u0022\u003EIn particular, in the LHS \u003Cstrong\u003E\u003Cem\u003E\u03b3\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ex\u003C\/em\u003E\u003C\/sub\u003E\u003Csup\u003E\u22a4\u003C\/sup\u003E \u00b7 \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E + \u003Cstrong\u003E\u003Cem\u003E\u03b3\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/sub\u003E\u003Csup\u003E\u22a4\u003C\/sup\u003E\u00b7\u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E of such a constraint, the decision \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003E\u03c4\u003C\/em\u003E\u003C\/sub\u003E, for \u003Cem\u003Ek\u003C\/em\u003E\u2264 \u003Cem\u003E\u03c4\u003C\/em\u003E\u2264 \u003Cem\u003Et\u003C\/em\u003E, will have coefficients\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-54\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-63.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-63.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nTo make calculations efficient, we note that the coefficients can be obtained recursively as in the C\u003Cspan class=\u0022sc\u0022\u003Ealculate\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eonstraint\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eoefficients\u003C\/span\u003E function defined in \u003Ca id=\u0022xref-statement-1-1\u0022 class=\u0022xref-statement\u0022 href=\u0022#statement-1\u0022\u003EAlgorithm 1\u003C\/a\u003E.\u003C\/p\u003E\u003Cdiv id=\u0022statement-1\u0022 class=\u0022statement\u0022\u003E\u003Cspan class=\u0022statement-label\u0022\u003EAlgorithm 1\u003C\/span\u003E\u003Ch3\u003EC\u003Cspan class=\u0022sc\u0022\u003Ealculate\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eonstraint\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eoefficients\u003C\/span\u003E\u003C\/h3\u003E\u003Cdiv id=\u0022F9\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F9.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F9.medium.gif\u0022 width=\u0022440\u0022 height=\u0022182\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F9.medium.gif\u0022 width=\u0022440\u0022 height=\u0022182\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F9.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure9\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F9.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-134\u0022\u003E\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-44\u0022\u003E\u003Ch3\u003EEC.3.3. Objective Coefficients\u003C\/h3\u003E\u003Cp id=\u0022p-135\u0022\u003EUp to constants, the objective in (16) can be written as\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-55\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-65.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-65.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwith\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-56\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-66.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-66.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nIn (16) the decisions \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u003C\/sub\u003E, for \u003Cem\u003Ek\u003C\/em\u003E \u2264 \u003Cem\u003Et\u003C\/em\u003E \u2264 \u003Cem\u003ET\u003C\/em\u003E \u2212 1, will have objective coefficients:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-57\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-67.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-67.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003C\/p\u003E\u003Cdiv id=\u0022statement-2\u0022 class=\u0022statement\u0022\u003E\u003Cspan class=\u0022statement-label\u0022\u003EAlgorithm 2\u003C\/span\u003E\u003Ch3\u003EC\u003Cspan class=\u0022sc\u0022\u003Ealculate\u003C\/span\u003EO\u003Cspan class=\u0022sc\u0022\u003Ebjective\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eoefficients\u003C\/span\u003E\u003C\/h3\u003E\u003Cdiv id=\u0022F10\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F10.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F10.medium.gif\u0022 width=\u0022440\u0022 height=\u0022356\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F10.medium.gif\u0022 width=\u0022440\u0022 height=\u0022356\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F10.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure10\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F10.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-136\u0022\u003E\u003C\/p\u003E\u003C\/div\u003E\u003Cp id=\u0022p-137\u0022\u003EThis allows calculating the coefficients recursively, just as we did for the constraints. The detailed function C\u003Cspan class=\u0022sc\u0022\u003Ealculate\u003C\/span\u003EO\u003Cspan class=\u0022sc\u0022\u003Ebjective\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eoefficients\u003C\/span\u003E is defined in \u003Ca id=\u0022xref-statement-2-1\u0022 class=\u0022xref-statement\u0022 href=\u0022#statement-2\u0022\u003EAlgorithm 2\u003C\/a\u003E.\u003C\/p\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-45\u0022\u003E\u003Ch4\u003ECalculation of \u003Cem\u003EM, \u03b3\u003C\/em\u003E and \u003Cem\u003E\u03b7\u003C\/em\u003E\u003C\/h4\u003E\u003Cp id=\u0022p-138\u0022\u003EExpanding the objective (11), we have:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-58\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-69.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-69.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nFrom this equation and the definitions of \u003Cem\u003Ev\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E (\u00b7) in Appendix (EC.2) above, we can write \u003Cstrong\u003E\u003Cem\u003EM\u003C\/em\u003E\u003C\/strong\u003E (where the rows are indexed by the controls and the columns indexed by the compartments) as\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-59\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-70.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-70.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-60\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-71.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-71.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nSimilarly, we can write \u003Cstrong\u003E\u003Cem\u003E\u03b3\u003C\/em\u003E\u003C\/strong\u003E (indexed by the compartments for each group) as\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-61\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-72.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-72.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-62\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-73.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-73.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-63\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-74.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-74.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nFinally, \u003Cstrong\u003E\u003Cem\u003E\u03b7\u003C\/em\u003E\u003C\/strong\u003E (indexed by the compartments for each group) is\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-64\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-75.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-75.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere the only non-zeros are in the indices corresponding to compartment \u003Cem\u003ED\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E of each group \u003Cem\u003Eg\u003C\/em\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-46\u0022\u003E\u003Ch3\u003EEC.3.4. Specifics of the Iterative Linearization-Optimization Procedure\u003C\/h3\u003E\u003Cp id=\u0022p-139\u0022\u003EHaving defined functions C\u003Cspan class=\u0022sc\u0022\u003Ealculate\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eonstraint\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eoefficients\u003C\/span\u003E and C\u003Cspan class=\u0022sc\u0022\u003Ealculate\u003C\/span\u003EO\u003Cspan class=\u0022sc\u0022\u003Ebjective\u003C\/span\u003EC\u003Cspan class=\u0022sc\u0022\u003Eoefficients\u003C\/span\u003E, the L\u003Cspan class=\u0022sc\u0022\u003Einearization\u003C\/span\u003E-O\u003Cspan class=\u0022sc\u0022\u003Eptimization\u003C\/span\u003E function which is the main subroutine of ROLD is described in \u003Ca id=\u0022xref-statement-3-1\u0022 class=\u0022xref-statement\u0022 href=\u0022#statement-3\u0022\u003EAlgorithm 3\u003C\/a\u003E. This function builds the linear approximation for the remaining trajectory of the system, and optimizes it via an LP in a trust region of an infinity-norm \u003Cem\u003E\u03f5\u003C\/em\u003E-ball around the initial nominal control \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-52\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-52.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-52.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. We denote that \u003Cem\u003E\u03f5\u003C\/em\u003E-ball by \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-53\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-53.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-53.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003Cdiv id=\u0022statement-3\u0022 class=\u0022statement\u0022\u003E\u003Cspan class=\u0022statement-label\u0022\u003EAlgorithm 3\u003C\/span\u003E\u003Ch3\u003EL\u003Cspan class=\u0022sc\u0022\u003Einearization\u003C\/span\u003E-O\u003Cspan class=\u0022sc\u0022\u003Eptimization\u003C\/span\u003E\u003C\/h3\u003E\u003Cdiv id=\u0022F11\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F11.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F11.medium.gif\u0022 width=\u0022440\u0022 height=\u002294\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F11.medium.gif\u0022 width=\u0022440\u0022 height=\u002294\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F11.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure11\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F11.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-140\u0022\u003E\u003C\/p\u003E\u003C\/div\u003E\u003Cp id=\u0022p-141\u0022\u003EThe full ROLD procedure is given in \u003Ca id=\u0022xref-statement-4-1\u0022 class=\u0022xref-statement\u0022 href=\u0022#statement-4\u0022\u003EAlgorithm 4\u003C\/a\u003E. Within each period \u003Cem\u003Ek\u003C\/em\u003E, ROLD calls the L\u003Cspan class=\u0022sc\u0022\u003Einearization\u003C\/span\u003E-O\u003Cspan class=\u0022sc\u0022\u003Eptimization\u003C\/span\u003E function iteratively up to a termination condition, using the output control to initialize the nominal control and the trust region for the next call of the function.\u003C\/p\u003E\u003Cp id=\u0022p-142\u0022\u003EThis still requires to choose an initialization of the \u003Cem\u003Ek\u003C\/em\u003E = 0 nominal control \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-54\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-54.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-54.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E; in our experiments we initialize this with a heuristic solution generated via a gradient method.\u003C\/p\u003E\u003Cp id=\u0022p-143\u0022\u003EFor the termination condition, we combine a fixed upper bound on the number of iterations with a condition that we do not repeat the control sequences produced by L\u003Cspan class=\u0022sc\u0022\u003Einearization\u003C\/span\u003E-O\u003Cspan class=\u0022sc\u0022\u003Eptimization\u003C\/span\u003E, in order to avoid cycles. The fixed upper bound on the number of iterations is set so as to ensure that for each \u003Cem\u003Ek\u003C\/em\u003E, every confinement decision in \u003Cstrong\u003E\u003Cem\u003Eu\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Ek\u003C\/em\u003E:\u003Cem\u003ET\u003C\/em\u003E \u22121\u003C\/sub\u003E can be changed to any value in [0, 1] with \u003Cem\u003E\u03f5\u003C\/em\u003E-length steps, i.e., the upper bound is at least \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-55\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-55.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-55.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. We further multiply the allowed number of iterations by a multiple \u003Cem\u003Emult\u003C\/em\u003E \u2265 1, fixing the upper bound to be \u003Cem\u003Emult\u003C\/em\u003E \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-56\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-56.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-56.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-144\u0022\u003EWe experimented with different values of \u003Cem\u003E\u03f5\u003C\/em\u003E between 0.01 and 0.5 and values of \u003Cem\u003Emult\u003C\/em\u003E between 1 and 5. As expected, lower values of \u003Cem\u003E\u03f5\u003C\/em\u003E resulted in a more stable and higher performing heuristic.\u003C\/p\u003E\u003Cp id=\u0022p-145\u0022\u003EHigher values of \u003Cem\u003Emult\u003C\/em\u003E improved the heuristic only up to around \u003Cem\u003Emult\u003C\/em\u003E = 2, after which point the non-cycling termination condition was triggered almost always. On the other hand, reducing \u003Cem\u003E\u03f5\u003C\/em\u003E had a significant impact on the run-time of the linearization algorithm. We chose the combination of \u003Cem\u003E\u03f5\u003C\/em\u003E and \u003Cem\u003Emult\u003C\/em\u003E that gave us the best trade-off between the quality of the solution and the total run-time. In particular, for all the runs presented we take \u003Cem\u003E\u03f5\u003C\/em\u003E = 0.05 and \u003Cem\u003Emult\u003C\/em\u003E = 2, resulting in an upper bound of 40 runs for the inner loop.\u003C\/p\u003E\u003Cdiv id=\u0022statement-4\u0022 class=\u0022statement\u0022\u003E\u003Cspan class=\u0022statement-label\u0022\u003EAlgorithm 4\u003C\/span\u003E\u003Ch3\u003EROLD\u003C\/h3\u003E\u003Cdiv id=\u0022F12\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F12.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F12.medium.gif\u0022 width=\u0022440\u0022 height=\u0022162\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F12.medium.gif\u0022 width=\u0022440\u0022 height=\u0022162\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F12.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure12\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F12.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-146\u0022\u003E\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-47\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EEC.4. Details on Parametrization and Calibration of the Model for \u00cele-de-France\u003C\/h2\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-48\u0022\u003E\u003Ch3\u003EEC.4.1. Basic SEIR Model Parameters\u003C\/h3\u003E\u003Cp id=\u0022p-147\u0022\u003EThe SEIR model parameters that are constant across age groups are summarized in \u003Ca id=\u0022xref-table-wrap-2-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T2\u0022\u003ETable EC.1\u003C\/a\u003E. The age-group specific parameters are reported in \u003Ca id=\u0022xref-table-wrap-3-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T3\u0022\u003ETable EC.2\u003C\/a\u003E. We start with the parameters as reported in \u003Ca id=\u0022xref-ref-44-4\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. (2020\u003C\/a\u003E)\u003Csup\u003E\u003Ca id=\u0022xref-fn-15-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-15\u0022\u003E10\u003C\/a\u003E\u003C\/sup\u003E, and then we allow the values of parameters \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-57\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-57.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-57.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E to change through time so as to model changes in how hospitals manage COVID-19 patients and changes in mandates for using masks and other measures that reduce transmission (details are presented in \u003Ca id=\u0022xref-sec-50-2\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-50\u0022\u003ESection EC.4.2\u003C\/a\u003E).\u003C\/p\u003E\u003Cp id=\u0022p-148\u0022\u003EFor \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E0\u003C\/sub\u003E and \u003Cem\u003E\u03bb\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003EH\u003C\/em\u003E\u003C\/sub\u003E, the reported uncertainty ranges are 95% confidence intervals. For \u003Cem\u003E\u03c3\u003C\/em\u003E\u003Csup\u003E\u22121\u003C\/sup\u003E (i.e., the mean stay in compartment \u003Cem\u003EE\u003C\/em\u003E), the uncertainty range is calculated as 4 \u00b1 0.8\u00b7 0.6 days, where 0.6 days is half the width of the 95% confidence interval for the incubation period reported in \u003Ca id=\u0022xref-ref-10-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-10\u0022\u003EBi et al. (2020\u003C\/a\u003E), and 0.8 accounts for the fact that the stay in compartment \u003Cem\u003EE\u003C\/em\u003E is 4\u003Cem\u003E\/\u003C\/em\u003E5 of the mean incubation time in \u003Ca id=\u0022xref-ref-44-5\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. (2020\u003C\/a\u003E). For \u003Cem\u003E\u00b5\u003C\/em\u003E\u003Csup\u003E\u22121\u003C\/sup\u003E (i.e., the mean stay in an infectious state), the uncertainty range is calculated as 4 \u00b1 0.43, where 0.43 is half the width of the 95% confidence interval for the serial interval reported by \u003Ca id=\u0022xref-ref-16-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-16\u0022\u003EDu et al. (2020\u003C\/a\u003E).\u003Csup\u003E\u003Ca id=\u0022xref-fn-16-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-16\u0022\u003E11\u003C\/a\u003E\u003C\/sup\u003E For the average stay in the ICU, we add to the mean stay of 20.46 days for \u00cele-de-France, another 1.5 day, which is the mean time spent in hospital prior to ICU admission (\u003Ca id=\u0022xref-ref-44-6\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. 2020\u003C\/a\u003E).\u003C\/p\u003E\u003Cdiv id=\u0022T2\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450877\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450877\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450877\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.1\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003ESEIR model parameters\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv id=\u0022T3\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450881\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450881\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450881\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.2\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EAge-group specific SEIR model probability parameters\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-49\u0022\u003E\u003Ch4\u003ECalculating the transmission rate \u003Cem\u003E\u03b2\u003C\/em\u003E from \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E0\u003C\/sub\u003E\u003C\/h4\u003E\u003Cp id=\u0022p-149\u0022\u003EWe obtain \u003Cem\u003E\u03b2\u003C\/em\u003E by linearizing the dynamics for \u003Cem\u003EE\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, \u003Cem\u003EI\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E around a point where \u003Cem\u003ES\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eh\u003C\/em\u003E\u003C\/sub\u003E \u2248 \u003Cem\u003EN\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eh\u003C\/em\u003E\u003C\/sub\u003E, \u003Cem\u003EI\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eh\u003C\/em\u003E\u003C\/sub\u003E \u2248 0, \u2200 \u003Cem\u003Eh\u003C\/em\u003E. More precisely, we have:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-65\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-80.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-80.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nThen, with \u003Cstrong\u003E\u003Cem\u003EY\u003C\/em\u003E\u003C\/strong\u003E (\u003Cem\u003Et\u003C\/em\u003E) :=[\u003Cem\u003EE\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E (\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003EE\u003C\/em\u003E\u003Csub\u003E2\u003C\/sub\u003E (\u003Cem\u003Et\u003C\/em\u003E), \u2026, \u003Cem\u003EE\u003C\/em\u003E\u003Csub\u003E|\ud835\udca2|\u003C\/sub\u003E, \u003Cem\u003EI\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E (\u003Cem\u003Et\u003C\/em\u003E), \u2026, \u003Cem\u003EI\u003C\/em\u003E\u003Csub\u003E|\ud835\udca2|\u003C\/sub\u003E]\u003Csup\u003E\u003Cem\u003ET\u003C\/em\u003E\u003C\/sup\u003E, we can write \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-58\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-58.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-58.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, where\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-66\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-81.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-81.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nand\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-67\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-82.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-82.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nThen \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E0\u003C\/sub\u003E can be identified as the spectral radius (i.e., the largest absolute value of the eigenvalues) of the matrix \u003Cstrong\u003E\u03a6\u039b\u003C\/strong\u003E\u003Csup\u003E\u22121\u003C\/sup\u003E (\u003Ca id=\u0022xref-ref-56-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-56\u0022\u003EDiekmann et al. 2010\u003C\/a\u003E, \u003Ca id=\u0022xref-ref-67-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-67\u0022\u003EPerasso 2018\u003C\/a\u003E). Since the eigenvalues of a matrix \u003Cem\u003E\u03b2\u003C\/em\u003E \u00b7\u003Cstrong\u003E\u003Cem\u003EA\u003C\/em\u003E\u003C\/strong\u003E are simply \u003Cem\u003E\u03b2\u003C\/em\u003E multiples of the eigenvalues of \u003Cstrong\u003E\u003Cem\u003EA\u003C\/em\u003E\u003C\/strong\u003E, we can therefore determine \u003Cem\u003E\u03b2\u003C\/em\u003E as \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E0\u003C\/sub\u003E divided by the spectral radius of the matrix (\u2212\u003Cstrong\u003E\u03a6\u003C\/strong\u003E\u003Cem\u003E\/\u03b2\u003C\/em\u003E) \u003Cstrong\u003E\u039b\u003C\/strong\u003E\u003Csup\u003E\u22121\u003C\/sup\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-50\u0022\u003E\u003Ch3\u003EEC.4.2. Epidemiological Model Parameter Fitting Using Health Outcomes and Mobility Data\u003C\/h3\u003E\u003Cp id=\u0022p-150\u0022\u003EWe use data on health outcomes from the French Public Health Agency (\u003Ca id=\u0022xref-ref-23-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-23\u0022\u003EFrench Government 2020\u003C\/a\u003E), as well as Google mobility data (\u003Ca id=\u0022xref-ref-27-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-27\u0022\u003EGoogle 2020\u003C\/a\u003E), to estimate the unknown parameters in our model. The data on health outcomes includes counts for individuals who are in the hospital, in the ICU, and who have died, by age group, and is maintained and updated daily by the French Public Health Agency (Sant\u00e9 publique France). The Google mobility data reports changes in activity at different places compared to a baseline, and is calculated using aggregate and anonymized data. For both health outcomes and mobility, we use data specific to the \u00cele-de-France region.\u003C\/p\u003E\u003Cp id=\u0022p-151\u0022\u003EThe calibration exercise has two purposes: (a) to further refine the SEIR parameters reported in the literature to the data observed in \u00cele-de-France and (b) to estimate our other parameters for which we do not have existing references. We then use the values of the estimated parameters in our experiments and simulations.\u003C\/p\u003E\u003Cp id=\u0022p-152\u0022\u003EWe first describe the set of parameters to be estimated, which we denote by \ud835\udcab.\u003C\/p\u003E\u003Cul class=\u0022list-unord \u0022 id=\u0022list-4\u0022\u003E\u003Cli id=\u0022list-item-11\u0022\u003E\u003Cp id=\u0022p-153\u0022\u003EDate of patient zero. We assume that the SEIR process starts with an infected individual of the 40-49 y.o. age group (\u003Ca id=\u0022xref-ref-66-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-66\u0022\u003EMohammad 2020\u003C\/a\u003E). We wish to estimate the date when this infection occurs.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-12\u0022\u003E\u003Cp id=\u0022p-154\u0022\u003EEpidemiological parameters. We use the epidemiological parameters of \u003Ca id=\u0022xref-ref-44-7\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. (2020\u003C\/a\u003E) to initialize the SEIR model. We allow these parameters to change in time in order to model changes in the way hospitals manage COVID-19 patients, as well as changes in mandates for using masks and other measures that reduce transmission. We assume that on date \u003Cem\u003Ed\u003C\/em\u003E, each parameter from the \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-59\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-59.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-59.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E set changes with respect to its initial value (as reported in \u003Ca id=\u0022xref-sec-48-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-48\u0022\u003ESection EC.4.1\u003C\/a\u003E), according to the relationship\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-68\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-83.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-83.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u003Cem\u003Em\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Es\u003C\/em\u003E\u003C\/sub\u003E is a multiplier pertaining to parameter \u003Cem\u003Es\u003C\/em\u003E. We assume the same multiplier \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-60\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-60.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-60.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E for all groups \u003Cem\u003Eg\u003C\/em\u003E \u2208 \ud835\udca2, and similarly for \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-61\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-61.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-61.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. We seek to determine the date of change \u003Cem\u003Ed\u003C\/em\u003E as well as the multipliers \u003Cem\u003Em\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Es\u003C\/em\u003E\u003C\/sub\u003E, \u003Cem\u003Es\u003C\/em\u003E\u2208\ud835\udcae.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-13\u0022\u003E\u003Cp id=\u0022p-155\u0022\u003EConfinement patterns. To estimate activity levels for the activities in our social mixing model, we use Google mobility data (\u003Ca id=\u0022xref-ref-27-4\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-27\u0022\u003EGoogle 2020\u003C\/a\u003E). The mobility data reports changes of activity (visits and length of stay) for each day, compared to a baseline value. The baseline used corresponds to the median value for the corresponding day of the week, during the five-week period January 3 -February 6 2020. We fix the \u003Ckbd\u003Ehome\u003C\/kbd\u003E activity level to be equal to 1, throughout time. We estimate the level of the other activities using the corresponding activities from the Google mobility data, as shown in \u003Ca id=\u0022xref-table-wrap-4-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T4\u0022\u003ETable EC.3\u003C\/a\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-156\u0022\u003EWhat remains to be estimated in the calibration is the weight parameter \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E\u003Ckbd\u003Eother\u003C\/kbd\u003E\u003C\/sub\u003E, as well as the school activity levels. We calibrate the level of schooling activity for four different time periods until October 21 2020. These periods are chosen to reflect (i) the dates when the French government closed down schools, and (ii) the French school calendar and summer recess.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-14\u0022\u003E\u003Cp id=\u0022p-157\u0022\u003ESocial mixing parameter. To reduce the number of parameters to be calibrated, we simplify our mixing model in (1) by constraining \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E = \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E2\u003C\/sub\u003E. We seek to determine this mixing parameter.\u003C\/p\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003Cdiv id=\u0022T4\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450879\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450879\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450879\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.3\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EMapping between the activities in our model and the activities in the Google mobility data\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-158\u0022\u003EWe next describe the details of the fitting procedure that we set up in order to retrieve an optimal parameter fitting. The mixing dynamics of the SEIR model are driven by the vector of activity levels \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E of each age group (as described in \u003Ca id=\u0022xref-sec-13-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-13\u0022\u003ESection 3.2\u003C\/a\u003E). Data on activity levels can be noisy; we model this uncertainty by assuming the vector of activity levels is a random vector, distributed as follows:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-69\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-85.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-85.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nThe value \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-62\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-62.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-62.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E is obtained from the Google activity data at time \u003Cem\u003Et\u003C\/em\u003E; this dataset does not differentiate activity by age, so \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-63\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-63.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-63.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E for all \u003Cem\u003Eg, h\u003C\/em\u003E \u2208 \ud835\udca2; in other words, all groups are assigned the same activity level. Recall that \u003Cem\u003EH\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003EICU\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003ED\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E) denote the hospital utilization, ICU utilization and cumulative number of deaths according to the SEIR model, respectively. We denote these quantities with \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-64\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-64.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-64.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E to emphasize the dependence of the SEIR process on parameters \ud835\udcab and the vector of activity levels \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-65\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-65.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-65.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. We aggregate these quantities over all age groups:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-70\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-86.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-86.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-71\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-87.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-87.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-72\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-88.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-88.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nDenote by \u003Cem\u003EH\u003C\/em\u003E\u003Csup\u003E\u003Cem\u003Eobs\u003C\/em\u003E\u003C\/sup\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003EICU\u003C\/em\u003E\u003Csup\u003E\u003Cem\u003Eobs\u003C\/em\u003E\u003C\/sup\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003ED\u003C\/em\u003E\u003Csup\u003E\u003Cem\u003Eobs\u003C\/em\u003E\u003C\/sup\u003E(\u003Cem\u003Et\u003C\/em\u003E) the general ward hospital beds utilization, ICU beds utilization, and cumulative deaths, respectively, at time \u003Cem\u003Et\u003C\/em\u003E, as observed in the real data for \u00cele-de-France from the French Public Health Agency (\u003Ca id=\u0022xref-ref-23-4\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-23\u0022\u003EFrench Government 2020\u003C\/a\u003E). We calculate the relative fitting error of the SEIR model at time \u003Cem\u003Et\u003C\/em\u003E for each of these three quantities as\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-73\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-89.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-89.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-74\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-90.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-90.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-75\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-91.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-91.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nWe define the total expected fitting error as a sum of these errors over different time intervals:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-76\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-92.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-92.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-77\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-93.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-93.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere the expectation is taken with respect to random vector \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-66\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-66.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-66.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, and where the time intervals are defined in \u003Ca id=\u0022xref-table-wrap-5-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T5\u0022\u003ETable EC.4\u003C\/a\u003E and comprise the entire period between March 17 2020 and October 21 2020. Our approach penalizes the errors at the peak times for hospital beds utilization, ICU beds utilization, and deaths; it also penalizes errors over the last 14 days of the considered period, to ensure an accurate fit at the end of the calibration horizon. We use different weights to account for the different errors. We use \u003Cem\u003Ew\u003C\/em\u003E\u003Csup\u003Epeak\u003C\/sup\u003E = 1\u003Cem\u003E\/\u003C\/em\u003E6 to account for the relatively smaller period of the peaks. We use \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-67\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-67.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-67.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E and\u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-68\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-68.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-68.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, with a higher weight for ICU beds utilization so as to ensure low error in the tail predictions of ICU utilization, as ICU beds utilization towards the tail of the calibration window plays an important role in the dynamics of the model right after the calibration window.\u003C\/p\u003E\u003Cdiv id=\u0022T5\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450880\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450880\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450880\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.4\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003ETime intervals used in the calibration\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-159\u0022\u003EWe seek to determine the set of parameters \ud835\udcab that minimize \u2130\u003Csup\u003E\ud835\udcab\u003C\/sup\u003E in (EC.53). We approximate the expectation in (EC.53) through a Monte Carlo sample-average approximation, using 100 samples. The set \ud835\udcab contains both discrete and continuous parameters. To minimize \u2130\u003Csup\u003E\ud835\udcab\u003C\/sup\u003E, we first do a grid search over all possible combinations for the discrete parameters, and then for each such combination, we perform a gradient descent procedure over the space of the continuous parameters. Each parameter is optimized within an allowed range, which is informed from the context of \u00cele-de-France in 2020. We used a wide allowed range if we had no information on what could be reasonable values for a given parameter.\u003C\/p\u003E\u003Cp id=\u0022p-160\u0022\u003EOur calibration procedure yields the parameter fitting summarized in \u003Ca id=\u0022xref-table-wrap-6-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T6\u0022\u003ETable EC.5\u003C\/a\u003E. We compare the fitted values of the SEIR model with the values reported by the French Public Health Agency in \u003Ca id=\u0022xref-fig-13-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F13\u0022\u003EFigure EC.1\u003C\/a\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-51\u0022\u003E\u003Ch3\u003EEC.4.3. Economic Model Parameter Fitting\u003C\/h3\u003E\u003Cp id=\u0022p-161\u0022\u003EWe obtain data on population, employment, and wages from the French National Institute of Statistics and Economic Studies (Institut national de la statistique et des \u00e9tudes \u00e9conomiques \u2014 INSEE). Where relevant, we discount all cash flows at a 3% annualized rate. We set the retirement age to be 65 (i.e., 64 is the last working year of age.)\u003C\/p\u003E\u003Cp id=\u0022p-162\u0022\u003EWe first obtain the initial population data \u003Cem\u003EN\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(0) for each age group in \u00cele-de-France at the end of 2019 from \u003Ca id=\u0022xref-ref-64-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-64\u0022\u003EINSEE (2020)\u003C\/a\u003E.\u003C\/p\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-52\u0022\u003E\u003Ch4\u003EEstimation of \u003Cem\u003Ew\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E\u003C\/h4\u003E\u003Cp id=\u0022p-163\u0022\u003ERecall that \u003Cem\u003Ew\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E in (EC.20) corresponds to the employment value for a member of group \u003Cem\u003Eg\u003C\/em\u003E, under normal conditions. To estimate \u003Cem\u003Ew\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E, we use two datasets from INSEE:\u003C\/p\u003E\u003Cdiv id=\u0022T6\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450892\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450892\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450892\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.5\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003ECalibration: fitted epidemiological and economic parameters and allowed estimation ranges\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv id=\u0022F13\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F13.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Predictions of the fitted SEIR model vs. reported values by the French Public Health Agency for three different quantities: (a) Hospital Beds Utilization, (b) ICU Beds Utilization and (c) Cumulative Deaths. \u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;span xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot; class=\u0026quot;caption-title\u0026quot;\u0026gt;Predictions of the fitted SEIR model vs. reported values by the French Public Health Agency for three different quantities: (a) Hospital Beds Utilization, (b) ICU Beds Utilization and (c) Cumulative Deaths.\u0026lt;\/span\u0026gt; \u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.1\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F13.medium.gif\u0022 width=\u0022440\u0022 height=\u0022103\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.1\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F13.medium.gif\u0022 width=\u0022440\u0022 height=\u0022103\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F13.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F13.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.1\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EPredictions of the fitted SEIR model vs. reported values by the French Public Health Agency for three different quantities: (a) Hospital Beds Utilization, (b) ICU Beds Utilization and (c) Cumulative Deaths.\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022list-unord \u0022 id=\u0022list-5\u0022\u003E\u003Cli id=\u0022list-item-15\u0022\u003E\u003Cp id=\u0022p-164\u0022\u003EYearly full time equivalent (FTE)\u003Csup\u003E\u003Ca id=\u0022xref-fn-17-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-17\u0022\u003E12\u003C\/a\u003E\u003C\/sup\u003E wages and employed population count for \u00cele-de-France in 2016, broken up into the age groups \u201cunder 26 years old\u201d, \u201c26 to 49 years old\u201d and \u201cmore than 50 years old\u201d (\u003Ca id=\u0022xref-ref-62-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-62\u0022\u003EINSEE 2016b\u003C\/a\u003E).\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-16\u0022\u003E\u003Cp id=\u0022p-165\u0022\u003EFTE employment rates across the entire economy for the fourth quarter of 2019, bucketed by age groups \u201c15 to 24 years old\u201d, \u201c25 to 49 years old\u201d, \u201c50 to 64 years old\u201d, and \u201c55 to 64 years old\u201d (\u003Ca id=\u0022xref-ref-63-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-63\u0022\u003EINSEE 2019\u003C\/a\u003E).\u003C\/p\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003Cp id=\u0022p-166\u0022\u003ESince we do not have a consolidated data source for economic data split by our exact age group definitions, we use the above datasets to interpolate values for \u003Cem\u003Ew\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E. At a high level, we derive wage curves across age ranges.\u003C\/p\u003E\u003Cp id=\u0022p-167\u0022\u003EWe next explain the general procedure, as well as the additional assumptions we have made for the interpolation. First, for the construction of wage curve by age bucket:\u003C\/p\u003E\u003Cul class=\u0022list-unord \u0022 id=\u0022list-6\u0022\u003E\u003Cli id=\u0022list-item-17\u0022\u003E\u003Cp id=\u0022p-168\u0022\u003EWe assume that the national level employment rates from \u003Ca id=\u0022xref-ref-63-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-63\u0022\u003EINSEE (2019)\u003C\/a\u003E are equal to those of the \u00cele-de-France region. Because the age bucketing for our age groups is finer than the age bucketing in the data, we use interpolation. Specifically, we fit a piece-wise linear model (consisting of three pieces) to the four employment rates reported for the \u201c15 to 24 years old\u201d, \u201c25 to 49 years old\u201d, \u201c50 to 64 years old\u201d, and \u201c55 to 64 years old\u201d groups. We take the midpoint of the age group as the \u003Cem\u003Ex\u003C\/em\u003E value of the datapoint; for example, for \u201c50 to 64 years old\u201d we use a midpoint of 57.5.\u003C\/p\u003E\u003Cp id=\u0022p-169\u0022\u003EWith this model, we can infer an employment rate for any arbitrary age and construct an employment rate curve.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-18\u0022\u003E\u003Cp id=\u0022p-170\u0022\u003EWe perform a similar fitting procedure for the age group wage information from \u003Ca id=\u0022xref-ref-62-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-62\u0022\u003EINSEE (2016b)\u003C\/a\u003E; since the wage progression by age is much smoother, we use simple linear regression to construct a wage curve for each one of our age buckets.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-19\u0022\u003E\u003Cp id=\u0022p-171\u0022\u003EThe previous wage curve only accounts for the employed population, whereas our age groups count the entire population. We thus combine the wage curve with employment rate and population data to arrive at a wage number blended across an entire age group\u2019s population.\u003C\/p\u003E\u003Cp id=\u0022p-172\u0022\u003EWhen doing this, we treat the 10-19 y.o. and 60-69 y.o. age groups specially by assuming the employment rates are reported only with respect to the work-eligible population in that bucket (15-19 and 60-64 year olds, respectively). We also set the work-eligible population for the 0-9, 70-79, and 80+ age buckets to 0. The formula we use is\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-78\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-97.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-97.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-20\u0022\u003E\u003Cp id=\u0022p-173\u0022\u003EThe interpolations we use introduce errors: in particular, if we aggregate the wages inferred by our constructed curve across the entire population, we overestimate the real total wages by 5.12%. We shade down all wages average_wage\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E proportionally so as to retrieve the real total wage amount \u003Cem\u003Ew\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E.\u003C\/p\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003Cp id=\u0022p-174\u0022\u003E\u003Ca id=\u0022xref-table-wrap-7-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T7\u0022\u003ETable EC.6\u003C\/a\u003E summarizes the year-based employment contribution parameters per age group. We note that when using them in the objective of the optimization problem, we divide these year-based values by 365, in order to capture employment value on a daily basis.\u003C\/p\u003E\u003Cdiv id=\u0022T7\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450895\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450895\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450895\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.6\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EYear-based economic value parameters under normal activity, per age group (in \u20ac)\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-53\u0022\u003E\u003Ch4\u003EEstimation of \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Ework\u003C\/kbd\u003E\u003C\/sup\u003E, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Efixed\u003C\/sup\u003E\u003C\/h4\u003E\u003Cp id=\u0022p-175\u0022\u003EWe move on to the estimation of parameters \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Ework\u003C\/kbd\u003E\u003C\/sup\u003E, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Efixed\u003C\/sup\u003E in (EC.20). These measure the sensitivity of economic value to the confinement pattern \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E(\u003Cem\u003Et\u003C\/em\u003E). We estimate them from data on lost economic output during the first lockdown phase employed in \u00cele-de-France, and in particular using the month of April 2020. We break up the approach into a few steps:\u003C\/p\u003E\u003Cul class=\u0022list-unord \u0022 id=\u0022list-7\u0022\u003E\u003Cli id=\u0022list-item-21\u0022\u003E\u003Cp id=\u0022p-176\u0022\u003EWe use survey data of French managers regarding business activity during the lockdown starting March 17 2020 from the Bank of France. This is sentiment data where managers are asked to compare current business conditions to normal conditions for the same relevant time period (\u003Ca id=\u0022xref-ref-53-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-53\u0022\u003EBanque de France 2020a\u003C\/a\u003E,b). These data are reported by industry, and we aggregate them into a single number weighting by industry size. We use FTE wages and employed population count for the \u00cele-de-France region in 2016 (\u003Ca id=\u0022xref-ref-61-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-61\u0022\u003EINSEE 2016a\u003C\/a\u003E) to figure out the appropriate weights to use in the aggregation. We then use these monthly readings as proxies for the economic activity level due to confinements in the month of April 2020, as compared to normal activity. The economic activity level for the month of April is 58.51%.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-22\u0022\u003E\u003Cp id=\u0022p-177\u0022\u003EA requirement for our estimation are the precise levels of confinement in April 2020. We retrieve these from Google mobility data (\u003Ca id=\u0022xref-ref-27-5\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-27\u0022\u003EGoogle 2020\u003C\/a\u003E), as explained in \u003Ca id=\u0022xref-sec-50-3\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-50\u0022\u003ESection EC.4.2\u003C\/a\u003E. To simplify the estimation, we set \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E = 0 and then determine parameters \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Ework\u003C\/kbd\u003E\u003C\/sup\u003E, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Efixed\u003C\/sup\u003E solving the system of equations\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-79\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-99.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-99.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-80\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-100.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-100.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-69\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-69.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-69.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E corresponds to the average value of \u003Cem\u003El\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Ework\u003C\/kbd\u003E\u003C\/sup\u003E(\u003Cem\u003Et\u003C\/em\u003E) through the month of April 2020. In our experiments, we also test our algorithm in alternative scenarios where we set \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E \u003Cem\u003E\u0026gt;\u003C\/em\u003E 0, keep the value for \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Efixed\u003C\/sup\u003E from the system (EC.55)-(EC.56), and adjust \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003E\u003Ckbd\u003Ework\u003C\/kbd\u003E\u003C\/sup\u003E = 1 \u2212 \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E \u2212 \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Efixed\u003C\/sup\u003E. The specific values we test are \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E \u2208 {0, 0.1, 0.2}.\u003C\/p\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-54\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EEC.5. Benchmark Policies\u003C\/h2\u003E\u003Cp id=\u0022p-178\u0022\u003EWe compare ROLD to several simpler classes of policies drawing inspiration by real life confinement management rules:\u003C\/p\u003E\u003Cul class=\u0022list-unord \u0022 id=\u0022list-8\u0022\u003E\u003Cli id=\u0022list-item-23\u0022\u003E\u003Cp id=\u0022p-179\u0022\u003E\u003Cstrong\u003EICU admissions trigger policy \u2014 ICU-t\u003C\/strong\u003E. This class of policies is similar in spirit to the trigger rule proposed by \u003Ca id=\u0022xref-ref-17-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-17\u0022\u003EDuque et al. (2020)\u003C\/a\u003E for the Austin metropolitan area. This rule places all age groups and activities (except \u003Ckbd\u003Ehome\u003C\/kbd\u003E) at a strict level of confinement when the average seven-day hospital admissions (i.e., inflow of patients admitted into the hospital due to COVID-19 complications) exceeds a pre-determined threshold, and then changes the confinement to a relaxed level when the average seven-day hospital admissions and the hospital utilization rate drop below pre-determined thresholds.\u003C\/p\u003E\u003Cp\u003ESince \u003Ca id=\u0022xref-ref-17-4\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-17\u0022\u003EDuque et al. (2020)\u003C\/a\u003E does not differentiate between hospital and ICU beds, we define our policy class on ICU admissions and utilization instead of hospitalizations. Specifically, the ICU admissions trigger policy is defined as in \u003Ca id=\u0022xref-statement-5-1\u0022 class=\u0022xref-statement\u0022 href=\u0022#statement-5\u0022\u003EAlgorithm 5\u003C\/a\u003E.\n\u003C\/p\u003E\u003Cdiv id=\u0022statement-5\u0022 class=\u0022statement\u0022\u003E\u003Cspan class=\u0022statement-label\u0022\u003EAlgorithm 5\u003C\/span\u003E\u003Ch3\u003EICU A\u003Cspan class=\u0022sc\u0022\u003Edmissions\u003C\/span\u003E T\u003Cspan class=\u0022sc\u0022\u003Erigger\u003C\/span\u003E P\u003Cspan class=\u0022sc\u0022\u003Eolicy\u003C\/span\u003E \u2014 ICU-t\u003C\/h3\u003E\u003Cdiv id=\u0022F14\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F14.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F14.medium.gif\u0022 width=\u0022440\u0022 height=\u0022151\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F14.medium.gif\u0022 width=\u0022440\u0022 height=\u0022151\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F14.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure14\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F14.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-181\u0022\u003E\u003C\/p\u003E\u003C\/div\u003E\u003Cp id=\u0022p-182\u0022\u003EWe optimize over the parameters \u003Cem\u003El\u003C\/em\u003E\u003Csub\u003Estrict\u003C\/sub\u003E, \u003Cem\u003El\u003C\/em\u003E\u003Csub\u003Erelaxed\u003C\/sub\u003E, \u003Cem\u003E\u03c1\u003C\/em\u003E\u003Csub\u003Eadmissions\u003C\/sub\u003E and \u003Cem\u003E\u03c1\u003C\/em\u003E\u003Csub\u003Eutilization\u003C\/sub\u003E via grid search with the goal of minimizing the objective in (11) corresponding to the total economic and death loss due to the pandemic, and we report the performance of the best policy.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-24\u0022\u003E\u003Cp id=\u0022p-183\u0022\u003E\u003Cstrong\u003EHybrid Trigger Policy \u2014 Hybrid-t\u003C\/strong\u003E. This policy resembles the rule used in France for declaring a region in \u201cmaximum alert\u201d\u003Csup\u003E\u003Ca id=\u0022xref-fn-18-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-18\u0022\u003E13\u003C\/a\u003E\u003C\/sup\u003E (\u003Ca id=\u0022xref-ref-32-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-32\u0022\u003ELehot and Borgne 2020\u003C\/a\u003E). Like the previous policy, this policy also switches between a (uniform) strict and a more relaxed confinement level, but the trigger condition combines ICU utilization with the rate of new infections in the population. Specifically, the policy switches to strict confinement if the average seven-day incidence rate in the population, defined\u003Csup\u003E\u003Ca id=\u0022xref-fn-19-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-19\u0022\u003E14\u003C\/a\u003E\u003C\/sup\u003E as \u2211\u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u22127\u2264\u003Cem\u003E\u03c4\u003C\/em\u003E \u2264\u003Cem\u003Et\u003C\/em\u003E\u22121\u003C\/sub\u003E \u2211 \u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E new infections\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003E\u03c4\u003C\/em\u003E)\u003Cem\u003E\/\u003C\/em\u003E \u2211 \u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E \u003Cem\u003EN\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(0), is greater than a threshold \u003Cem\u003E\u03c1\u003C\/em\u003E\u003Csub\u003Eincidence\u003C\/sub\u003E, \u003Cem\u003Eand\u003C\/em\u003E the incidence rate in age groups corresponding to the population that is 60 y.o. and above, \u2211 \u003Csub\u003E\u003Cem\u003Et\u003C\/em\u003E\u22127\u2264\u003Cem\u003E\u03c4\u003C\/em\u003E \u2264\u003Cem\u003Et\u003C\/em\u003E\u22121\u003C\/sub\u003E \u2211 \u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u226560 y.o.\u003C\/sub\u003E new infections\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003E\u03c4\u003C\/em\u003E)\u003Cem\u003E\/\u003C\/em\u003E \u2211 \u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u226560 y.o.\u003C\/sub\u003E \u003Cem\u003EN\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(0), is greater than a threshold \u003Cem\u003E\u03c1\u003C\/em\u003E\u003Csub\u003Eincidence_60+\u003C\/sub\u003E, \u003Cem\u003Eand\u003C\/em\u003E ICU utilization rate is greater than a threshold \u003Cem\u003E\u03c1\u003C\/em\u003E\u003Csub\u003Eutilization\u003C\/sub\u003E. We optimize over all parameters \u003Cem\u003El\u003C\/em\u003E\u003Csub\u003Estrict\u003C\/sub\u003E, \u003Cem\u003El\u003C\/em\u003E\u003Csub\u003Erelaxed\u003C\/sub\u003E, \u003Cem\u003E\u03c1\u003C\/em\u003E\u003Csub\u003Eincidence\u003C\/sub\u003E, \u003Cem\u003E\u03c1\u003C\/em\u003E\u003Csub\u003Eincidence_60+\u003C\/sub\u003E and \u003Cem\u003E\u03c1\u003C\/em\u003E\u003Csub\u003Eutilization\u003C\/sub\u003E with the goal of minimizing the total loss objective in (11), and we report the performance of the best policy.\u003C\/p\u003E\u003Cp\u003EThis is the Hybrid-t AND policy, and it is described in \u003Ca id=\u0022xref-statement-6-1\u0022 class=\u0022xref-statement\u0022 href=\u0022#statement-6\u0022\u003EAlgorithm 6\u003C\/a\u003E. We also test a stricter version of this policy that takes the logical \u003Cem\u003Eor\u003C\/em\u003E of the three conditions (Hybrid-t OR), instead of taking the \u003Cem\u003Eand\u003C\/em\u003E, as the trigger for setting the strict confinement level.\n\u003C\/p\u003E\u003Cdiv id=\u0022statement-6\u0022 class=\u0022statement\u0022\u003E\u003Cspan class=\u0022statement-label\u0022\u003EAlgorithm 6\u003C\/span\u003E\u003Ch3\u003EH\u003Cspan class=\u0022sc\u0022\u003Eybrid\u003C\/span\u003E T\u003Cspan class=\u0022sc\u0022\u003Erigger\u003C\/span\u003E P\u003Cspan class=\u0022sc\u0022\u003Eolicy\u003C\/span\u003E \u2014 Hybrid-t AND\u003C\/h3\u003E\u003Cdiv id=\u0022F15\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F15.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F15.medium.gif\u0022 width=\u0022440\u0022 height=\u0022152\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F15.medium.gif\u0022 width=\u0022440\u0022 height=\u0022152\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F15.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure15\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F15.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-185\u0022\u003E\u003C\/p\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-25\u0022\u003E\u003Cp id=\u0022p-186\u0022\u003E\u003Cstrong\u003EFully open \u2014 FO\u003C\/strong\u003E. This corresponds to the normal conditions where \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-70\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-70.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-70.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E for all \u003Cem\u003Ea\u003C\/em\u003E \u2208 {\u003Ckbd\u003Ehome\u003C\/kbd\u003E, \u003Ckbd\u003Ework\u003C\/kbd\u003E, \u003Ckbd\u003Eschool\u003C\/kbd\u003E, \u003Ckbd\u003Etransport\u003C\/kbd\u003E, \u003Ckbd\u003Eleisure\u003C\/kbd\u003E, \u003Ckbd\u003Eother\u003C\/kbd\u003E}, \u003Cem\u003Eg\u003C\/em\u003E \u2208 \ud835\udca2 and \u003Cem\u003Et\u003C\/em\u003E \u2208 {0, \u2026, \u003Cem\u003ET\u003C\/em\u003E \u2212 1}.\u003C\/p\u003E\u003C\/li\u003E\u003Cli id=\u0022list-item-26\u0022\u003E\u003Cp id=\u0022p-187\u0022\u003E\u003Cstrong\u003EFull confinement \u2014 FC\u003C\/strong\u003E. In this policy, all activities except \u003Ckbd\u003Ehome\u003C\/kbd\u003E are fully restricted. That is, we set \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-71\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-71.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-71.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E for all \u003Cem\u003Ea\u003C\/em\u003E \u2208 {\u003Ckbd\u003Ework\u003C\/kbd\u003E, \u003Ckbd\u003Eschool\u003C\/kbd\u003E, \u003Ckbd\u003Etransport\u003C\/kbd\u003E, \u003Ckbd\u003Eleisure\u003C\/kbd\u003E, \u003Ckbd\u003Eother\u003C\/kbd\u003E}, \u003Cem\u003Eg\u003C\/em\u003E \u2208 \ud835\udca2, \u003Cem\u003Et\u003C\/em\u003E \u2208 {0, \u2026, \u003Cem\u003ET\u003C\/em\u003E \u2212 1}, and \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-72\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-72.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-72.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E for all \u003Cem\u003Eg\u003C\/em\u003E \u2208 \ud835\udca2 and \u003Cem\u003Et\u003C\/em\u003E \u2208 {0, \u2026, \u003Cem\u003ET\u003C\/em\u003E \u2212 1}.\u003C\/p\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-55\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EEC.6. Additional Results\u003C\/h2\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-56\u0022\u003E\u003Ch3\u003EEC.6.1. Robustness Analyses\u003C\/h3\u003E\u003Cp id=\u0022p-188\u0022\u003EWe analyze additional problem instances by changing the value of each of 13 estimated parameters within a sensitivity range, as shown in \u003Ca id=\u0022xref-table-wrap-8-1\u0022 class=\u0022xref-table\u0022 href=\u0022#T8\u0022\u003ETable EC.7\u003C\/a\u003E. For each parameter, we sample 40 values uniformly at random from a specified sensitivity range. In each problem instance, one parameter is changed from its estimated value, for a total of 13 \u00d7 40 = 520 problem instances.\u003C\/p\u003E\u003Cdiv id=\u0022T8\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450898\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450898\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450898\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.7\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003ERobustness analysis: parameters and sensitivity analysis ranges\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-189\u0022\u003E\u003Ca id=\u0022xref-fig-16-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F16\u0022\u003EFigure EC.2\u003C\/a\u003E shows robustness results for seven values of the economic cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E: [0, 10, 15, 25, 50, 100, 150] \u00d7 the annual GDP per capita in France. The shown boxplots summarize results over the 520 problem instances, for each value of \u003Cem\u003E\u03c7\u003C\/em\u003E. These results reinforce our findings from \u003Ca id=\u0022xref-sec-27-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-27\u0022\u003ESection 6.1\u003C\/a\u003E on the gains of dual targeting, as well as the observation that dual targeting unlocks complementarities which may not be available under targeting age groups or activities separately.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-57\u0022\u003E\u003Ch3\u003EEC.6.2. Can Dual Targeting Reduce Time in Confinement for Each Age Group?\u003C\/h3\u003E\u003Cp id=\u0022p-190\u0022\u003EWe calculate the fraction of time each age group spends in confinement under each ROLD policy, averaged over the activities relevant to that age group. Specifically, these are defined as:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-81\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-104.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-104.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-82\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-105.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-105.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-83\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-106.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-106.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-84\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-107.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-107.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003C\/p\u003E\u003Cdiv id=\u0022F16\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F16.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Robustness analyses showing the superiority of ROLD AGE-ACT over ROLD policies with less granular targeting, and the super-additive improvements of ROLD AGE-ACT over the sum of the improvements of ROLD AGE and ROLD ACT, for different values for the cost of death \u0026#x3C7;, over a wide set of problem instances. All improvements are with respect to ROLD NO-TARGET. For each value of \u0026#x3C7;, the boxplots summarize results over 520 different problem instances.\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;\u0026lt;strong\u0026gt;Robustness analyses showing the superiority of ROLD AGE-ACT over ROLD policies with less granular targeting, and the super-additive improvements of ROLD AGE-ACT over the sum of the improvements of ROLD AGE and ROLD ACT, for different values for the cost of death\u0026lt;\/strong\u0026gt; \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt;, \u0026lt;strong\u0026gt;over a wide set of problem instances. All improvements are with respect to ROLD NO-TARGET. For each value of\u0026lt;\/strong\u0026gt; \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt;, \u0026lt;strong\u0026gt;the boxplots summarize results over 520 different problem instances\u0026lt;\/strong\u0026gt;.\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.2\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F16.medium.gif\u0022 width=\u0022440\u0022 height=\u0022168\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.2\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F16.medium.gif\u0022 width=\u0022440\u0022 height=\u0022168\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F16.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.2\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F16.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.2\u003C\/span\u003E \u003Cp id=\u0022p-191\u0022 class=\u0022first-child\u0022\u003E\u003Cstrong\u003ERobustness analyses showing the superiority of ROLD AGE-ACT over ROLD policies with less granular targeting, and the super-additive improvements of ROLD AGE-ACT over the sum of the improvements of ROLD AGE and ROLD ACT, for different values for the cost of death\u003C\/strong\u003E \u003Cem\u003E\u03c7\u003C\/em\u003E, \u003Cstrong\u003Eover a wide set of problem instances. All improvements are with respect to ROLD NO-TARGET. For each value of\u003C\/strong\u003E \u003Cem\u003E\u03c7\u003C\/em\u003E, \u003Cstrong\u003Ethe boxplots summarize results over 520 different problem instances\u003C\/strong\u003E.\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-192\u0022\u003EWe make some further comments about the results shown in \u003Ca id=\u0022xref-fig-5-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F5\u0022\u003EFigure 5\u003C\/a\u003E in the main paper. It is worth noting that although ROLD AGE-ACT reduces confinements for every age group compared to less targeted policies, it does not do so uniformly, and it can lead to an even larger discrepancy in the confinements faced by different age groups: those aged 10-59 are generally more confined than those aged 0-9 or 60+. Such disparate treatment is non-existent in the NO-TARGET policy by definition, and is less apparent in the AGE and ACT policies. Interestingly, while AGE generally exhibits a similar discrimination profile to AGE-ACT, ACT exhibits an almost opposite profile, with more confinement applied to those aged 70+.\u003C\/p\u003E\u003Cp id=\u0022p-193\u0022\u003EThis is explainable by considering the econ-to-contacts-ratio (explained in \u003Ca id=\u0022xref-sec-28-1\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-28\u0022\u003ESection 6.2\u003C\/a\u003E): for instance, since \u003Ckbd\u003Eleisure\u003C\/kbd\u003E is responsible for the smallest (population-weighted average) econ-to-contacts-ratio among all activities, it is the first activity that ACT targets for confinement, and this disproportionately impacts the groups aged 70+, who are active in fewer activities overall than other age groups and have \u003Ckbd\u003Eleisure\u003C\/kbd\u003E as one of their activities.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-58\u0022\u003E\u003Ch3\u003EEC.6.3. Additional Details for the Regression Decision Trees for the ROLD Policy\u003C\/h3\u003E\u003Cp id=\u0022p-194\u0022\u003ETo understand how the ROLD policies target confinements across different age groups and activities, we train an interpretable machine learning model \u2013 a regression decision tree \u2013 to predict the optimal ROLD confinement in each activity as a function of several features.\u003C\/p\u003E\u003Cp id=\u0022p-195\u0022\u003ETo build a training set, we first create a larger set of problem instances built using a wide range of problem parameters. First, we create a set of 125 samples of tuplets of transmission rate \u003Cem\u003E\u03b2\u003C\/em\u003E, probabilities of developing severe symptoms conditional on infection {\u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Ess,g\u003C\/em\u003E\u003C\/sub\u003E }\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E and social mixing parameters which we set to equal values as \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E12\u003C\/sub\u003E := \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E1\u003C\/sub\u003E = \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E2\u003C\/sub\u003E. We create each sample by drawing each of these three parameters independently, according to the following distributions. For \u003Cem\u003E\u03b2\u003C\/em\u003E, we consider the range for \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E0\u003C\/sub\u003E of [2.8, 2.99], which is the 95% confidence interval from \u003Ca id=\u0022xref-ref-44-8\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. (2020\u003C\/a\u003E) (\u003Ca id=\u0022xref-table-wrap-2-2\u0022 class=\u0022xref-table\u0022 href=\u0022#T2\u0022\u003ETable EC.1\u003C\/a\u003E), and then sample a normal with mean 2.9 and standard deviation 0.0475 (such that the [2.8, 2.99] interval is covered by four standard deviations), truncating it if the value lies beyond the above lower and upper bounds for \u003Cem\u003ER\u003C\/em\u003E\u003Csub\u003E0\u003C\/sub\u003E. We proceed similarly for{\u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Ess,g\u003C\/em\u003E\u003C\/sub\u003E} \u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E, except that we keep the proportions of these probabilities across age groups constant. Specifically, we create ranges for each \u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Ess,g\u003C\/em\u003E\u003C\/sub\u003E from the \u003Ca id=\u0022xref-ref-44-9\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. (2020\u003C\/a\u003E) 95% confidence intervals (\u003Ca id=\u0022xref-table-wrap-3-2\u0022 class=\u0022xref-table\u0022 href=\u0022#T3\u0022\u003ETable EC.2\u003C\/a\u003E) but sample a a common standard normal which we then scale to generate individual \u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Ess,g\u003C\/em\u003E\u003C\/sub\u003E\u2019s. We use the same procedure for \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E12\u003C\/sub\u003E except that we use the range [0.1, 0.7] for this parameter.\u003C\/p\u003E\u003Cp id=\u0022p-196\u0022\u003EWe also vary the parameters \u003Cem\u003E\u03c7\u003C\/em\u003E \u2208 {0, 10, 20, \u2026, 90, 100, 150, 250, \u2026, 950}, \u003Cem\u003E\u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E \u2208 {0, 0.1, 0.2} and \u003Cem\u003EK\u003C\/em\u003E\u003Csub\u003EICU\u003C\/sub\u003E \u2208 {2600, 2900, 3200}. Specifically, for each of the 125 (\u003Cem\u003E\u03b2\u003C\/em\u003E, {\u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Ess,g\u003C\/em\u003E\u003C\/sub\u003E}\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u2208\ud835\udca2\u003C\/sub\u003E, \u003Cem\u003E\u03b1\u003C\/em\u003E\u003Csub\u003E12\u003C\/sub\u003E) tuplets thus sampled, we try every possible combination of \u003Cem\u003E\u03c7, \u03bd\u003C\/em\u003E\u003Csup\u003Eother activities\u003C\/sup\u003E, \u003Cem\u003EK\u003C\/em\u003E\u003Csup\u003EICU\u003C\/sup\u003E in the above discrete ranges. All resulting instances use a horizon of \u003Cem\u003ET\u003C\/em\u003E = 90 days and change the confinement decisions every 14 days. We also add the 3640 instances we generate for robustness analysis (Appendix EC.6.1) to our ensemble of instances.\u003C\/p\u003E\u003Cp id=\u0022p-197\u0022\u003EFor each problem instance, we compute the optimal ROLD activity levels, i.e., the decisions \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-73\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-73.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-73.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E for \u003Cem\u003Eg\u003C\/em\u003E\u2208 \ud835\udca2, \u003Cem\u003Ea\u003C\/em\u003E\u2208 \ud835\udc9c, 0\u2264 \u003Cem\u003Et\u003C\/em\u003E\u2264 \u003Cem\u003ET\u003C\/em\u003E\u22121, which we then use to simulate the SEIR dynamics and calculate all the corresponding values of the SEIR states. Lastly, we create a training set of 8,232,210 samples \u2014 with one sample for every 14-day period (i.e., for \u003Cem\u003Et\u003C\/em\u003E = 0, 14, 28, etc.), each age group \u003Cem\u003Eg\u003C\/em\u003E, and each activity \u003Cem\u003Ea\u003C\/em\u003E relevant to that age group \u2014 where we include several direct and derived features based on the parameter values characterizing the instance and the induced SEIR states, and a target corresponding to the optimal ROLD decision \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-74\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-74.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-74.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-198\u0022\u003EWe consider both targeted features, by which we mean features that differentiate on either group \u003Cem\u003Eg\u003C\/em\u003E or activity \u003Cem\u003Ea\u003C\/em\u003E,\u003Csup\u003E\u003Ca id=\u0022xref-fn-20-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-20\u0022\u003E15\u003C\/a\u003E\u003C\/sup\u003E or on both, and non-targeted features (\u003Ca id=\u0022xref-table-wrap-9-2\u0022 class=\u0022xref-table\u0022 href=\u0022#T9\u0022\u003ETables EC.8\u003C\/a\u003E, \u003Ca id=\u0022xref-table-wrap-10-2\u0022 class=\u0022xref-table\u0022 href=\u0022#T10\u0022\u003EEC.9\u003C\/a\u003E). Some of the non-targeted features are allowed to depend on \u003Cem\u003Et\u003C\/em\u003E.\u003C\/p\u003E\u003Cdiv id=\u0022T9\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450872\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450872\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450872\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.8\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003ETargeted features used for tree fitting\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-199\u0022\u003EUsing this data, we then use the CART algorithm (implemented in the scikit-learn Python library) to train regression decision trees to predict the optimal ROLD confinement decisions across all activities and age groups as a function of the considered features, using the traditional mean-squared-error (MSE) criterion as a goodness of fit metric. When using \u003Cem\u003Ek\u003C\/em\u003E-fold cross-validation to assess the optimal depth of the tree, we noticed that optimal depths exceeded 10. As such, we decided to train a tree of depth four for interpretability purposes, and we also trained trees with maximum depth ranging from four to ten to test the robustness of our results.\u003C\/p\u003E\u003Cp id=\u0022p-200\u0022\u003EIn addition, we also report in \u003Ca id=\u0022xref-fig-17-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F17\u0022\u003EFigure EC.3\u003C\/a\u003E the tree of depth five trained using \u003Cem\u003Eall\u003C\/em\u003E the features described above. Beyond appearing in the first four levels of the tree (which is consistent with \u003Ca id=\u0022xref-fig-7-3\u0022 class=\u0022xref-fig\u0022 href=\u0022#F7\u0022\u003EFigure 7\u003C\/a\u003E), it is worth noting that the econ-to-contacts-ratio continues to be used for splits in the fifth level, which further supports the importance of that feature.\u003C\/p\u003E\u003Cp id=\u0022p-201\u0022\u003ELastly, we perform a few other tests of feature importance. \u003Ca id=\u0022xref-fig-18-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F18\u0022\u003EFigure EC.4\u003C\/a\u003E reports permutation importance for a very high depth (equal to 10) tree. While at this higher depth there are more features with non-zero permutation importance, econ-to-contacts-ratio continues to overwhelmingly dominate the others.\u003C\/p\u003E\u003Cdiv id=\u0022T10\u0022 class=\u0022table pos-float\u0022\u003E\u003Cdiv class=\u0022table-inline table-callout-links\u0022\u003E\u003Cdiv class=\u0022callout\u0022\u003E\u003Cspan\u003EView this table:\u003C\/span\u003E\u003Cul class=\u0022callout-links\u0022\u003E\u003Cli class=\u0022view-inline first\u0022\u003E\u003Ca href=\u0022\u0022 class=\u0022table-expand-inline\u0022 data-table-url=\u0022\/highwire\/markup\/450869\/expansion?postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0026amp;table-expand-inline=1\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView inline\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022view-popup\u0022\u003E\u003Ca href=\u0022\/highwire\/markup\/450869\/expansion?width=1000\u0026amp;height=500\u0026amp;iframe=true\u0026amp;postprocessors=highwire_tables%2Chighwire_reclass%2Chighwire_figures%2Chighwire_math%2Chighwire_inline_linked_media%2Chighwire_embed\u0022 class=\u0022colorbox colorbox-load table-expand-popup\u0022 rel=\u0022gallery-fragment-tables\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EView popup\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022download-ppt last\u0022\u003E\u003Ca href=\u0022\/highwire\/powerpoint\/450869\u0022 class=\u0022highwire-figure-link highwire-figure-link-ppt\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload powerpoint\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022table-caption\u0022\u003E\u003Cspan class=\u0022table-label\u0022\u003ETable EC.9\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003ENon-targeted features used for tree fitting\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-202\u0022\u003EWe also train another set of depth four trees to verify this finding through a different means than permutation feature importance. Specifically, we train a tree with \u003Cem\u003Eonly\u003C\/em\u003E the non-targeted features and trees with (all) the non-targeted features plus a single targeted feature. We do this to see which targeted feature most improves goodness-of-fit versus the non-targeted tree. \u003Ca id=\u0022xref-fig-19-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F19\u0022\u003EFigure EC.5\u003C\/a\u003E shows the econ-to-contacts-ratio clearly dominates the other trees, and is very close in RMSE to the tree using all features reported in \u003Ca id=\u0022xref-fig-7-4\u0022 class=\u0022xref-fig\u0022 href=\u0022#F7\u0022\u003EFigure 7\u003C\/a\u003E.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-59\u0022\u003E\u003Ch3\u003EEC.6.4. More Details on the Behavior of the ROLD Policy\u003C\/h3\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-60\u0022\u003E\u003Ch4\u003EConfinements and testing\u003C\/h4\u003E\u003Cp id=\u0022p-203\u0022\u003E\u003Ca id=\u0022xref-fig-20-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F20\u0022\u003EFigure EC.6a\u003C\/a\u003E depicts the outcomes and actions of an optimal ROLD policy that is able to target both confinement and testing decisions over a 90-day optimization horizon, for a cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u00d7 and under a daily testing capacity of 60,000 tests. ROLD cleverly mixes confinement decisions and targeted testing. For example, it distributes the testing capacity to the 40-69 age groups, who are also kept open in \u003Ckbd\u003Ework\u003C\/kbd\u003E. Intuitively, this is because the \u003Ckbd\u003Ework\u003C\/kbd\u003E activity generates more social contacts (and thus new infections) than other activities. Moreover, of all the age groups (10-69) that are active in \u003Ckbd\u003Ework\u003C\/kbd\u003E, although the 40-69s do not have the largest number of social contacts, they do have the most contacts with the high risk 70+ age groups, so most of the testing capacity is focused on them. In other words, the tests are used to detect and quarantine the individuals which are most likely to create new infections in high risk individuals through their social contacts in open activities.\u003C\/p\u003E\u003Cp id=\u0022p-204\u0022\u003EAdding in testing can substantially change the confinement policy by permitting less total confinement of the population. When comparing with \u003Ca id=\u0022xref-fig-6-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F6\u0022\u003EFigure 6\u003C\/a\u003E, \u003Ca id=\u0022xref-fig-20-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F20\u0022\u003EFigure EC.6a\u003C\/a\u003E shows that the ability to test reduces the overall confinement in \u003Ckbd\u003Ework\u003C\/kbd\u003E, \u003Ckbd\u003Eleisure\u003C\/kbd\u003E and \u003Ckbd\u003Eother\u003C\/kbd\u003E. And although testing is mostly concentrated in the 40-49 y.o. and 50-59 y.o. groups, it also serves to reduce confinement in some groups which are not tested, such as the confinement in work for the 10-19 y.o. group.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-61\u0022\u003E\u003Ch4\u003EDependence of the proposed policy on problem parameters\u003C\/h4\u003E\u003Cp id=\u0022p-205\u0022\u003EROLD decisions generally depend quite strongly on the specific problem parameters, such as the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E. We make a few other observations on how the ROLD policies depend on critical problem parameters.\u003C\/p\u003E\u003Cdiv id=\u0022F17\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F17.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Decision tree of depth five approximating the optimized ROLD confinement decisions (trained with 8,232,210 samples), with a horizon of T = 90 days. \u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;\u0026lt;span class=\u0026quot;caption-title\u0026quot;\u0026gt;Decision tree of depth five approximating the optimized ROLD confinement decisions (trained with 8,232,210 samples), with a horizon of \u0026lt;em\u0026gt;T\u0026lt;\/em\u0026gt; = 90 days.\u0026lt;\/span\u0026gt; \u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.3\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F17.medium.gif\u0022 width=\u0022340\u0022 height=\u0022440\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.3\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F17.medium.gif\u0022 width=\u0022340\u0022 height=\u0022440\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F17.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.3\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F17.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.3\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EDecision tree of depth five approximating the optimized ROLD confinement decisions (trained with 8,232,210 samples), with a horizon of \u003Cem\u003ET\u003C\/em\u003E = 90 days.\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-62\u0022\u003E\u003Ch5\u003ECost of death \u03c7\u003C\/h5\u003E\u003Cp id=\u0022p-206\u0022\u003EAt any fixed time, as \u003Cem\u003E\u03c7\u003C\/em\u003E increases there is \u003Cem\u003Egenerally\u003C\/em\u003E a nestedness in the confinement decisions followed by ROLD: if a certain age group is confined in a particular activity at a fixed time for a value of \u003Cem\u003E\u03c7\u003C\/em\u003E, that group will face a confinement that is at least as strict in that activity, at that time, for a larger value of \u003Cem\u003E\u03c7\u003C\/em\u003E. As a comparison to the policy for \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u00d7, \u003Ca id=\u0022xref-fig-20-3\u0022 class=\u0022xref-fig\u0022 href=\u0022#F20\u0022\u003EFigure EC.6b\u003C\/a\u003E reports the decisions produced by ROLD for a cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E = 150\u00d7 the annual\u003C\/p\u003E\u003Cp id=\u0022p-207\u0022\u003EGDP per capita, using an optimization horizon of \u003Cem\u003ET\u003C\/em\u003E = 90 days. This property can be very useful for decision making, especially in cases where a policy maker cannot or prefers not to assign a specific value for \u003Cem\u003E\u03c7\u003C\/em\u003E, but instead is comfortable providing a range for \u003Cem\u003E\u03c7\u003C\/em\u003E. It is important to note that this property of nested confinements does not immediately follow, e.g., as \u003Cem\u003E\u03c7\u003C\/em\u003E increases and a policy decides to impose more strict confinements on specific age groups, it could relax the confinements of other groups.\u003C\/p\u003E\u003Cdiv id=\u0022F18\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F18.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Permutation feature importance for different variables, for the depth 10 decision tree. \u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;span xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot; class=\u0026quot;caption-title\u0026quot;\u0026gt;Permutation feature importance for different variables, for the depth 10 decision tree.\u0026lt;\/span\u0026gt; \u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.4\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F18.medium.gif\u0022 width=\u0022440\u0022 height=\u0022206\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.4\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F18.medium.gif\u0022 width=\u0022440\u0022 height=\u0022206\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F18.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.4\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F18.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.4\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EPermutation feature importance for different variables, for the depth 10 decision tree.\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv id=\u0022F19\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F19.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Percent decrease in RMSE for the single targeted feature trees and the all features tree, versus the tree using only non-targeted features. \u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;span xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot; class=\u0026quot;caption-title\u0026quot;\u0026gt;Percent decrease in RMSE for the single targeted feature trees and the all features tree, versus the tree using only non-targeted features.\u0026lt;\/span\u0026gt; \u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.5\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F19.medium.gif\u0022 width=\u0022440\u0022 height=\u0022223\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.5\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F19.medium.gif\u0022 width=\u0022440\u0022 height=\u0022223\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F19.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.5\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F19.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.5\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EPercent decrease in RMSE for the single targeted feature trees and the all features tree, versus the tree using only non-targeted features.\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-63\u0022\u003E\u003Ch5\u003ETime horizon T\u003C\/h5\u003E\u003Cp id=\u0022p-208\u0022\u003EIn order to understand the effect of the time horizon, we also run experiments for an optimization horizon of \u003Cem\u003ET\u003C\/em\u003E = 360 days. These reinforce the behavior observed for \u003Cem\u003ET\u003C\/em\u003E = 90 days. \u003Ca id=\u0022xref-fig-21-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F21\u0022\u003EFigure EC.7\u003C\/a\u003E reports the decisions produced by ROLD for an optimization horizon of \u003Cem\u003ET\u003C\/em\u003E = 360 days, using a cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u00d7 the annual GDP per capita. Comparing this with the policy for \u003Cem\u003ET\u003C\/em\u003E = 90 days, we see that the confinement policy for \u003Cem\u003ET\u003C\/em\u003E = 360 days is not monotonic over time. However, the policy still relaxes confinements as we get close to the end of the horizon. Furthermore, the confinement for a fixed (age group, activity) pair and time \u003Cem\u003Et\u003C\/em\u003E will generally be at least as strict as the optimization horizon increases. Exceptions to this general \u201crule\u201d can occur at early times of a long optimization horizon.\u003C\/p\u003E\u003Cdiv id=\u0022F20\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F20.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022The optimized ROLD confinement and testing decisions and their impact on hospitalizations, infections, and deaths, for different values of the cost of death, for a 90-day optimization horizon, with a testing capacity of 60,000 daily tests. \u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;span xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot; class=\u0026quot;caption-title\u0026quot;\u0026gt;The optimized ROLD confinement and testing decisions and their impact on hospitalizations, infections, and deaths, for different values of the cost of death, for a 90-day optimization horizon, with a testing capacity of 60,000 daily tests.\u0026lt;\/span\u0026gt; \u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.6\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F20.medium.gif\u0022 width=\u0022440\u0022 height=\u0022316\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.6\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F20.medium.gif\u0022 width=\u0022440\u0022 height=\u0022316\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F20.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.6\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F20.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.6\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EThe optimized ROLD confinement and testing decisions and their impact on hospitalizations, infections, and deaths, for different values of the cost of death, for a 90-day optimization horizon, with a testing capacity of 60,000 daily tests.\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-64\u0022\u003E\u003Ch3\u003EEC.6.5. Value of Targeted Testing and Quarantining\u003C\/h3\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-65\u0022\u003E\u003Ch4\u003EValue of targeting testing\u003C\/h4\u003E\u003Cp id=\u0022p-209\u0022\u003EWe next assess the value that can be generated by targeting testing based on age groups, with subsequent quarantining of positive cases. We compare the performance of two testing policies, while keeping all activities open, and varying the testing capacity: targeting testing to different age groups using ROLD, versus allocating testing proportionally to the census population of each age group. When we run experiments with viral testing, we consider a baseline capacity of 60,000 tests daily for \u00cele-de-France (\u003Ca id=\u0022xref-ref-52-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-52\u0022\u003EAgence R\u00e9gionale de Sant\u00e9 \u2014 \u00cele-de-France 2020\u003C\/a\u003E, reported on May 11, 2020).\u003C\/p\u003E\u003Cp id=\u0022p-210\u0022\u003E\u003Ca id=\u0022xref-fig-22-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F22\u0022\u003EFigure EC.8\u003C\/a\u003E reports the performance of these two testing policies. \u003Ca id=\u0022xref-fig-23-1\u0022 class=\u0022xref-fig\u0022 href=\u0022#F23\u0022\u003EFigure EC.9\u003C\/a\u003E reports the difference in deaths, economic losses, and total losses between targeted testing and proportional testing.\u003C\/p\u003E\u003Cp id=\u0022p-211\u0022\u003EAt a cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u00d7 the annual GDP per capita, the value of targeted testing with a capacity of 60,000 daily tests is an improvement by EUR 2.9B (7.8%) in terms of total losses (1,488 or 7.8% less deaths; EUR 0.1B or 8.3% less economic loss). The improvement in deaths and in the total loss due to targeting testing would be maximized (3,210 or 26.7% less deaths, EUR 6.3B or 26.9% less total losses) with a daily testing capacity of around 238,700 tests, corresponding to more than 1.9% of the \u00cele-de-France population. This is more than four times larger than the real testing capacity in \u00cele-de-France (\u003Ca id=\u0022xref-ref-69-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-69\u0022\u003ESterl\u00e9 2020\u003C\/a\u003E). \u003Ca id=\u0022xref-fig-22-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F22\u0022\u003EFigures EC.8\u003C\/a\u003E and \u003Ca id=\u0022xref-fig-23-2\u0022 class=\u0022xref-fig\u0022 href=\u0022#F23\u0022\u003EEC.9\u003C\/a\u003E report results for a wide range of testing capacities.\u003C\/p\u003E\u003Cdiv id=\u0022F21\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F21.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022The optimized ROLD confinement and testing decisions and their impact on hospitalizations, infections, and deaths for a 360-day optimization horizon, in a problem instance with cost of death \u0026#x3C7; = 50\u0026#xD7; and a testing capacity of 60,000 daily tests. \u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;\u0026lt;span class=\u0026quot;caption-title\u0026quot;\u0026gt;The optimized ROLD confinement and testing decisions and their impact on hospitalizations, infections, and deaths for a 360-day optimization horizon, in a problem instance with cost of death \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt; = 50\u0026lt;em\u0026gt;\u0026#xD7;\u0026lt;\/em\u0026gt; and a testing capacity of 60,000 daily tests.\u0026lt;\/span\u0026gt; \u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.7\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F21.medium.gif\u0022 width=\u0022440\u0022 height=\u0022418\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.7\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F21.medium.gif\u0022 width=\u0022440\u0022 height=\u0022418\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F21.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.7\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F21.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.7\u003C\/span\u003E \u003Cspan class=\u0022caption-title\u0022\u003EThe optimized ROLD confinement and testing decisions and their impact on hospitalizations, infections, and deaths for a 360-day optimization horizon, in a problem instance with cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u003Cem\u003E\u00d7\u003C\/em\u003E and a testing capacity of 60,000 daily tests.\u003C\/span\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cp id=\u0022p-212\u0022\u003EWe note that for targeted testing, the improvement in total losses from higher testing capacity starts slowing down around 230,000 daily tests, as shown in \u003Ca id=\u0022xref-fig-22-3\u0022 class=\u0022xref-fig\u0022 href=\u0022#F22\u0022\u003EFigure EC.8\u003C\/a\u003E; that\u2019s also the testing capacity at which the difference between targeted and proportional testing peaks, as shown in \u003Ca id=\u0022xref-fig-23-3\u0022 class=\u0022xref-fig\u0022 href=\u0022#F23\u0022\u003EFigure EC.9\u003C\/a\u003E. At that testing capacity, there is no more infection in the community. There are still benefits from a larger testing capacity, as the point in time at which the infection is killed would come sooner, however the benefits are diminishing.\u003C\/p\u003E\u003Cp id=\u0022p-213\u0022\u003EFor proportional testing, the improvement in total losses from higher testing capacity starts slowing down around 400,000 daily tests, as shown in \u003Ca id=\u0022xref-fig-22-4\u0022 class=\u0022xref-fig\u0022 href=\u0022#F22\u0022\u003EFigure EC.8\u003C\/a\u003E. First, it makes sense that the proportional testing curve would start slowing down for a larger testing capacity than the targeted testing curve. Second, that is also the testing capacity at which the drop in the difference between targeted and proportional testing tapers off, as shown in \u003Ca id=\u0022xref-fig-23-4\u0022 class=\u0022xref-fig\u0022 href=\u0022#F23\u0022\u003EFigure EC.9\u003C\/a\u003E. At large testing capacities, for which the infection is killed both for targeted testing and for proportional testing, targeted testing is still able to extract more value from abundant testing capacity than proportional testing.\u003C\/p\u003E\u003Cdiv id=\u0022F22\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F22.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Comparison of the performance of testing-and-quarantining-only interventions (with ROLD targeted testing vs. with proportional testing), and confinement-only interventions (with ROLD confinement decisions), for different testing capacities and different levels of effective confinement time. The experiments are run using \u0026#x3C7; = 50\u0026#xD7; the annual GDP per capita and an ICU capacity of 2900 beds.\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;\u0026lt;strong\u0026gt;Comparison of the performance of testing-and-quarantining-only interventions (with ROLD targeted testing vs. with proportional testing), and confinement-only interventions (with ROLD confinement decisions), for different testing capacities and different levels of effective confinement time. The experiments are run using\u0026lt;\/strong\u0026gt; \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt; = 50\u0026lt;em\u0026gt;\u0026#xD7;\u0026lt;\/em\u0026gt; \u0026lt;strong\u0026gt;the annual GDP per capita and an ICU capacity of 2900 beds\u0026lt;\/strong\u0026gt;.\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.8\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F22.medium.gif\u0022 width=\u0022440\u0022 height=\u0022263\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.8\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F22.medium.gif\u0022 width=\u0022440\u0022 height=\u0022263\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F22.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.8\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F22.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.8\u003C\/span\u003E \u003Cp id=\u0022p-214\u0022 class=\u0022first-child\u0022\u003E\u003Cstrong\u003EComparison of the performance of testing-and-quarantining-only interventions (with ROLD targeted testing vs. with proportional testing), and confinement-only interventions (with ROLD confinement decisions), for different testing capacities and different levels of effective confinement time. The experiments are run using\u003C\/strong\u003E \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u003Cem\u003E\u00d7\u003C\/em\u003E \u003Cstrong\u003Ethe annual GDP per capita and an ICU capacity of 2900 beds\u003C\/strong\u003E.\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv id=\u0022F23\u0022 class=\u0022fig pos-float type-figure  odd\u0022\u003E\u003Cdiv class=\u0022highwire-figure\u0022\u003E\u003Cdiv class=\u0022fig-inline-img-wrapper\u0022\u003E\u003Cdiv class=\u0022fig-inline-img\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F23.large.jpg?width=800\u0026amp;height=600\u0026amp;carousel=1\u0022 title=\u0022Comparison of targeted testing using ROLD versus proportional allocation of testing to age groups, in terms of deaths, economic losses, and total losses. The experiments are run while keeping all activities open, and using \u0026#x3C7; = 50\u0026#xD7; the annual GDP per capita and an ICU capacity of 2900 beds.\u0022 class=\u0022highwire-fragment fragment-images colorbox-load\u0022 rel=\u0022gallery-fragment-images-1079962778\u0022 data-figure-caption=\u0022\u0026lt;div class=\u0026quot;highwire-markup\u0026quot;\u0026gt;\u0026lt;div xmlns=\u0026quot;http:\/\/www.w3.org\/1999\/xhtml\u0026quot;\u0026gt;\u0026lt;strong\u0026gt;Comparison of targeted testing using ROLD versus proportional allocation of testing to age groups, in terms of deaths, economic losses, and total losses. The experiments are run while keeping all activities open, and using\u0026lt;\/strong\u0026gt; \u0026lt;em\u0026gt;\u0026#x3C7;\u0026lt;\/em\u0026gt; = 50\u0026lt;em\u0026gt;\u0026#xD7;\u0026lt;\/em\u0026gt; \u0026lt;strong\u0026gt;the annual GDP per capita and an ICU capacity of 2900 beds\u0026lt;\/strong\u0026gt;.\u0026lt;\/div\u0026gt;\u0026lt;\/div\u0026gt;\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003E\u003Cspan class=\u0022hw-responsive-img\u0022\u003E\u003Cimg class=\u0022highwire-fragment fragment-image lazyload\u0022 alt=\u0022Figure EC.9\u0022 src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F23.medium.gif\u0022 width=\u0022440\u0022 height=\u0022109\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-fragment fragment-image\u0022 alt=\u0022Figure EC.9\u0022 src=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F23.medium.gif\u0022 width=\u0022440\u0022 height=\u0022109\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cul class=\u0022highwire-figure-links inline\u0022\u003E\u003Cli class=\u0022download-fig first\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F23.large.jpg?download=true\u0022 class=\u0022highwire-figure-link highwire-figure-link-download\u0022 title=\u0022Download Figure EC.9\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EDownload figure\u003C\/a\u003E\u003C\/li\u003E\u003Cli class=\u0022new-tab last\u0022\u003E\u003Ca href=\u0022https:\/\/www.medrxiv.org\/content\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/F23.large.jpg\u0022 class=\u0022highwire-figure-link highwire-figure-link-newtab\u0022 target=\u0022_blank\u0022 data-icon-position=\u0022\u0022 data-hide-link-title=\u00220\u0022\u003EOpen in new tab\u003C\/a\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022fig-caption\u0022\u003E\u003Cspan class=\u0022fig-label\u0022\u003EFigure EC.9\u003C\/span\u003E \u003Cp id=\u0022p-215\u0022 class=\u0022first-child\u0022\u003E\u003Cstrong\u003EComparison of targeted testing using ROLD versus proportional allocation of testing to age groups, in terms of deaths, economic losses, and total losses. The experiments are run while keeping all activities open, and using\u003C\/strong\u003E \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u003Cem\u003E\u00d7\u003C\/em\u003E \u003Cstrong\u003Ethe annual GDP per capita and an ICU capacity of 2900 beds\u003C\/strong\u003E.\u003C\/p\u003E\u003Cdiv class=\u0022sb-div caption-clear\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-66\u0022\u003E\u003Ch4\u003EValue of confinement versus value of testing\u003C\/h4\u003E\u003Cp id=\u0022p-216\u0022\u003EWe also use our framework to compare (a) an intervention that only uses confinement and no testing to (b) an intervention that only uses testing and subsequent quarantining while keeping all activities open. To better understand the trade-offs, we parametrize each of these based on the most salient parameter controlling its respective performance: the overall level of confinement allowed for (a) and the testing capacity for (b).\u003C\/p\u003E\u003Cp id=\u0022p-217\u0022\u003EMore precisely, to implement (a) we constrain ROLD to use at most a specified \u201cbudget\u201d of total effective time in confinement by adding the following constraint to the optimization model:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-85\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-118.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-118.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003Elock\u003C\/sub\u003E is the maximum percentage of time that can be spent in confinement. We provide results for a range of values of \u003Cem\u003Ep\u003C\/em\u003E\u003Csub\u003Elock\u003C\/sub\u003E \u2208 [0, 1]. To implement (b), we constrain ROLD to use a specified testing capacity, and allow for optimized allocation of tests to different age groups. We run ROLD for a wide range of daily testing capacities \u003Cem\u003EK\u003C\/em\u003E\u003Csup\u003EVtest\u003C\/sup\u003E, from zero to 10% of the population.\u003C\/p\u003E\u003Cp id=\u0022p-218\u0022\u003EIn both cases we allow ROLD to target its decisions to different age groups. Specifically, for the confinement-only policy we do not allow ROLD to target activities, so as to keep the comparison between the confinement and the testing interventions fair: both are only able to target age groups.\u003C\/p\u003E\u003Cp id=\u0022p-219\u0022\u003E\u003Ca id=\u0022xref-fig-22-5\u0022 class=\u0022xref-fig\u0022 href=\u0022#F22\u0022\u003EFigure EC.8\u003C\/a\u003E reports the comparison in performance (in terms of the total loss objective) between the confinement-only and testing-only interventions, for a cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E = 50\u00d7 the annual GDP per capita. The plot shows that all the benefits from confinement can be accomplished with a total effective confinement time that corresponds to 19.2% of the total horizon window, or 17.3 days of effective confinement time within a 90-day horizon, as long as the confinement is optimally targeted to different age groups. To attain the same total loss (EUR 8.6B) without confinement, an extraordinary amount of (targeted) testing would be needed: more than 632,500 tests would need to be administered daily, which corresponds to more than 5.1% of the \u00cele-de-France population. The plot shows that testing without confinement can in theory reach performance levels that are unattainable by confinement without testing, however this only happens when the daily testing capacity exceeds 5.1% of the \u00cele-de-France population. This is about eleven times larger than the real testing capacity in \u00cele-de-France (\u003Ca id=\u0022xref-ref-69-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-69\u0022\u003ESterl\u00e9 2020\u003C\/a\u003E).\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-67\u0022\u003E\u003Ch3\u003EEC.6.6. Limited Disparity Requirements\u003C\/h3\u003E\u003Cp id=\u0022p-220\u0022\u003EWe impose limited disparity constraints on confinement decisions across pairs of age groups, for a given activity. For \u003Ckbd\u003Ework\u003C\/kbd\u003E, we impose constraints for age groups 10-19 y.o., \u2026, 60-69 y.o. For \u003Ckbd\u003Etransport\u003C\/kbd\u003E, \u003Ckbd\u003Eleisure\u003C\/kbd\u003E and \u003Ckbd\u003Eother\u003C\/kbd\u003E, we impose constraints for all age groups. No constraints are imposed for \u003Ckbd\u003Ehome\u003C\/kbd\u003E and for \u003Ckbd\u003Eschool\u003C\/kbd\u003E. Given a disparity tolerance \u0394 \u2265 0, we formalize the constraints as follows:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-86\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-119.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-119.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nSmaller values of \u0394 imply more stringent non-disparity constraints. We provide numerical results for a range of values of \u0394 in [0, 1].\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-68\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EEC.7. Stylized Model of Linearized Dynamics\u003C\/h2\u003E\u003Cp id=\u0022p-221\u0022\u003EThis section generates further insight into the structure of the confinement policies produced by ROLD by examining a significantly simpler version of the problem we study, for which an analytical characterization is possible. Our goal is to understand how the optimal ROLD policy depends on various problem primitives (such as economic parameters, contact matrices, etc.), which in turn will confirm our selection of features for training the tree policies in \u003Ca id=\u0022xref-sec-28-2\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-28\u0022\u003ESection 6.2\u003C\/a\u003E and the prominence of econ-to-contacts-ratio.\u003C\/p\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-69\u0022\u003E\u003Ch3\u003ESimplified SEIR model\u003C\/h3\u003E\u003Cp id=\u0022p-222\u0022\u003EWe consider a simplified compartmental model in which there is a single population group engaging in a single activity, and there are only susceptible (\u003Cem\u003ES\u003C\/em\u003E), exposed (\u003Cem\u003EE\u003C\/em\u003E), infectious (\u003Cem\u003EI\u003C\/em\u003E), recovered (\u003Cem\u003ER\u003C\/em\u003E) and deceased (\u003Cem\u003ED\u003C\/em\u003E) compartments. Also for the sake of simplicity, we consider the \u003Cem\u003Econtinuous\u003C\/em\u003E time dynamics of this model, namely:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-87\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-120.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-120.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-88\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-121.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-121.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-89\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-122.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-122.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-90\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-123.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-123.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-91\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-124.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-124.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nWe pause to explain the new dynamical equations. Since there is a single group and activity, the new infections term \u003Cem\u003E\u03b2Cl\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E)\u003Cem\u003ES\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E)\u003Cem\u003EI\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) is significantly simplified. Here, \u003Cem\u003E\u03b2\u003C\/em\u003E is the transmission rate and \u003Cem\u003EC\u003C\/em\u003E is the rate at which social contacts occur. Furthermore, \u003Cem\u003E\u03c3\u003C\/em\u003E and \u003Cem\u003E\u00b5\u003C\/em\u003E are the transition rates defined analogously to our original model. Since in this stylized model we remove hospitalized states, we have a direct transition from \u003Cem\u003EI\u003C\/em\u003E to \u003Cem\u003ER\u003C\/em\u003E or \u003Cem\u003ED\u003C\/em\u003E; we denote by \u003Cem\u003Ep\u003C\/em\u003E the probability that an individual dies given that they are infectious. We denote by \u003Cem\u003EN\u003C\/em\u003E the total population size.\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-70\u0022\u003E\u003Ch3\u003EActivity levels\u003C\/h3\u003E\u003Cp id=\u0022p-223\u0022\u003EThe control is the activity level of the population and is one-dimensional, so we denote it by \u003Cem\u003El\u003C\/em\u003E, in analogy to the original model. We simplify the economic model by taking the economic value as a linear function of the activity level \u003Cem\u003El\u003C\/em\u003E, i.e., \u003Cem\u003Ewl\u003C\/em\u003E, where \u003Cem\u003Ew\u003C\/em\u003E is the economic value generated per capita and per unit time under no confinement. Given the same cost of death parameter \u003Cem\u003E\u03c7\u003C\/em\u003E, the objective is to maximize:\u003Csup\u003E\u003Ca id=\u0022xref-fn-21-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-21\u0022\u003E16\u003C\/a\u003E\u003C\/sup\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-92\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-125.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-125.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003C\/p\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section\u0022 id=\u0022sec-71\u0022\u003E\u003Ch3\u003ESolving the linearized system\u003C\/h3\u003E\u003Cp id=\u0022p-224\u0022\u003ETo facilitate the analysis, we make the assumption that throughout the time horizon and for any activity level, \u003Cem\u003ES\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) \u2248 \u003Cem\u003EN\u003C\/em\u003E. In that case, the new infections term can be approximated as\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-93\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-126.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-126.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nleading to the approximate dynamics:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-94\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-127.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-127.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-95\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-128.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-128.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-96\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-129.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-129.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-97\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-130.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-130.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-98\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-131.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-131.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nIn compact notation, these approximate SEIR dynamics can be written as \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-75\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-75.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-75.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, where \u003Cstrong\u003E\u003Cem\u003EX\u003C\/em\u003E\u003C\/strong\u003E(\u003Cem\u003Et\u003C\/em\u003E) = (\u003Cem\u003ES\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003EE\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003EI\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003ER\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cem\u003ED\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E)) and the function \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-76\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-76.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-76.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003Ecaptures (EC.68)-(EC.72).\u003C\/p\u003E\u003Cp id=\u0022p-225\u0022\u003EAdditionally, under the assumption \u003Cem\u003ES\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) \u2248 \u003Cem\u003EN\u003C\/em\u003E, the objective simplifies to:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-99\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-132.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-132.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nsince \u003Cem\u003ES\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003EE\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003EI\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003ER\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) \u2248 \u003Cem\u003EN\u003C\/em\u003E.\u003C\/p\u003E\u003Cp id=\u0022p-226\u0022\u003EWith respect to this linearized dynamic and approximate objective, the control problem is:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-100\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-133.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-133.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-101\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-134.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-134.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-102\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-135.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-135.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nNote that the control problem under the simplified model does not include capacity constraints, as we have removed the hospitalized states from the model.\u003C\/p\u003E\u003Cp id=\u0022p-227\u0022\u003EAlthough the dynamics in (EC.68) - (EC.72) are linear, the bilinear terms coming from multiplying \u003Cem\u003El\u003C\/em\u003E with the SEIR states still make the problem difficult to solve. We thus proceed with a linearization of the problem which mimics the workings of the ROLD algorithm in \u003Ca id=\u0022xref-sec-19-2\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-19\u0022\u003ESection 4\u003C\/a\u003E, and characterize the optimal policy for this linearized problem. Analogously to \u003Ca id=\u0022xref-sec-19-3\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-19\u0022\u003ESection 4\u003C\/a\u003E, consider a nominal time-invariant control \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-77\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-77.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-77.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, and a nominal trajectory \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-78\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-78.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-78.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E. We build a linear approximation of the system dynamics as:\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-103\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-136.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-136.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-104\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-137.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-137.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-105\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-138.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-138.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nThis allows us to express the state at any \u003Cem\u003Et\u003C\/em\u003E as a function of the confinement decision \u003Cem\u003El\u003C\/em\u003E(\u003Cem\u003E\u03c4\u003C\/em\u003E) for all 0 \u2264 \u003Cem\u003E\u03c4\u003C\/em\u003E \u2264 \u003Cem\u003Et\u003C\/em\u003E. In particular, the solution of the dynamical system \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-79\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-79.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-79.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E is\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-106\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-139.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-139.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nNote that we can readily write the objective (EC.74) as maximizing\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-107\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-140.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-140.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwithout resorting to any Taylor approximations for linearization.\u003C\/p\u003E\u003Cp id=\u0022p-228\u0022\u003EBy plugging in the solution for the dynamical system (EC.80), we have\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-108\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-141.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-141.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-109\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-142.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-142.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-110\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-143.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-143.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-111\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-144.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-144.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere \u003Cstrong\u003E\u003Cem\u003Ee\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E3\u003C\/sub\u003E := [0, 0, 1, 0, 0].\u003C\/p\u003E\u003Cp id=\u0022p-229\u0022\u003EWe can now rewrite the objective in (EC.81) as\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-112\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-145.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-145.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nThe second summand is a constant with respect to the control. The coefficient of \u003Cem\u003El\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) in the integral in the first summand calculates to\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-113\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-146.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-146.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nand the optimal policy is to set \u003Cem\u003El\u003C\/em\u003E to 1 if and only if the expression above is non-negative.\u003C\/p\u003E\u003Cp id=\u0022p-230\u0022\u003EIt is perhaps most useful to focus on understanding the resulting policy for the case that \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-80\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-80.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-80.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E, in which case the coefficient of \u003Cem\u003El\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) takes a simpler form equal to\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-114\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-147.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-147.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nThe optimal decision then is\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-115\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-148.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-148.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nwhere\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-116\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-149.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-149.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\n(We note that\u003Csup\u003E\u003Ca id=\u0022xref-fn-22-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-22\u0022\u003E17\u003C\/a\u003E\u003C\/sup\u003E \u003Cem\u003Ef\u003C\/em\u003E (\u003Cem\u003E\u03b2, \u03c7, p, \u03c3, \u00b5, N, \u00ce, t, T\u003C\/em\u003E) \u22650 for 0\u2264 \u003Cem\u003Et\u003C\/em\u003E\u2264 \u003Cem\u003ET\u003C\/em\u003E.)\u003C\/p\u003E\u003Cp id=\u0022p-231\u0022\u003EInequality (EC.89) uncovers a natural logic for the confinement decisions in this linearized formulation. Specifically, the left-hand side exactly corresponds to the econ-to-contacts-ratio that we identified in \u003Ca id=\u0022xref-sec-28-3\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-28\u0022\u003ESection 6.2\u003C\/a\u003E: it is given by the gradient of (per-capita) economic value generated with respect to the level of activity \u003Cem\u003El\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E)\u003Csup\u003E\u003Ca id=\u0022xref-fn-23-1\u0022 class=\u0022xref-fn\u0022 href=\u0022#fn-23\u0022\u003E18\u003C\/a\u003E\u003C\/sup\u003E divided by the rate of social contacts generated. Thus, the optimal policy is governed by a threshold on the value of the econ-to-contacts-ratio, allowing normal activity levels (\u003Cem\u003El\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) = 1) when the econ-to-contacts-ratio exceeds the threshold and completely confining the entire population (\u003Cem\u003El\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) = 0) otherwise. The threshold, given by the function \u003Cem\u003Ef\u003C\/em\u003E (\u003Cem\u003E\u03b2, \u03c7, p, \u03c3, \u00b5, N, \u00ce, t, T\u003C\/em\u003E), is increasing in parameters such as the probability of infection given a contact \u003Cem\u003E\u03b2\u003C\/em\u003E, the probability of death given infection \u003Cem\u003Ep\u003C\/em\u003E, and the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E, and is decreasing in the size of the overall population \u003Cem\u003EN\u003C\/em\u003E, which matches intuition.\u003C\/p\u003E\u003Cp id=\u0022p-232\u0022\u003EWe remark that the simple threshold policy relying on econ-to-contacts-ratio emerges when we impose \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-81\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-81.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-81.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E in (EC.87); for general \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-82\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-82.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-82.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E,the policy still follows a threshold rule based on the marginal economic value exceeding a function that depends on problem parameters and the rate of contacts, but the simple ratio no longer emerges in the rule.\u003C\/p\u003E\u003Cp id=\u0022p-233\u0022\u003EThere are two conclusions we highlight from this stylized model analysis. First, even with significant simplifications such as a single group and activity, the policies output by ROLD or similar procedures are quite complex and elusive to completely characterize in closed form. Second, there are however interesting regimes where these policies intuitively depend on a quantity resembling the econ-to-contacts-ratio. In this light, selecting the group-activity pairs to be confined in decreasing order of their econ-to-contacts-ratios, as our trees from \u003Ca id=\u0022xref-sec-28-4\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-28\u0022\u003ESection 6.2\u003C\/a\u003E do, naturally parallels the policy in (EC.89). This provides some theoretical backing to the feature that the trees consider to be important in explaining ROLD\u2019s decisions.\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section ack\u0022 id=\u0022ack-1\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EAcknowledgments\u003C\/h2\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section fn-group\u0022 id=\u0022fn-group-1\u0022\u003E\u003Ch2\u003EFootnotes\u003C\/h2\u003E\u003Cul\u003E\u003Cli class=\u0022fn-others\u0022 id=\u0022fn-1\u0022\u003E\u003Cp id=\u0022p-1\u0022\u003E\u003Cspan class=\u0022em-link\u0022\u003E\u003Cspan class=\u0022em-addr\u0022\u003Ecamelo{at}stanford.edu\u003C\/span\u003E\u003C\/span\u003E,\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn-others\u0022 id=\u0022fn-2\u0022\u003E\u003Cp id=\u0022p-2\u0022\u003E\u003Cspan class=\u0022em-link\u0022\u003E\u003Cspan class=\u0022em-addr\u0022\u003Eflorin.ciocan{at}insead.edu\u003C\/span\u003E\u003C\/span\u003E,\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn-others\u0022 id=\u0022fn-3\u0022\u003E\u003Cp id=\u0022p-3\u0022\u003E\u003Cspan class=\u0022em-link\u0022\u003E\u003Cspan class=\u0022em-addr\u0022\u003Edaniancu{at}stanford.edu\u003C\/span\u003E\u003C\/span\u003E,\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn-others\u0022 id=\u0022fn-4\u0022\u003E\u003Cp id=\u0022p-4\u0022\u003E\u003Cspan class=\u0022em-link\u0022\u003E\u003Cspan class=\u0022em-addr\u0022\u003Exwarnes{at}stanford.edu\u003C\/span\u003E\u003C\/span\u003E,\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn-others\u0022 id=\u0022fn-5\u0022\u003E\u003Cp id=\u0022p-5\u0022\u003E\u003Cspan class=\u0022em-link\u0022\u003E\u003Cspan class=\u0022em-addr\u0022\u003Espyros.zoumpoulis{at}insead.edu\u003C\/span\u003E\u003C\/span\u003E,\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-6\u0022\u003E\u003Cp id=\u0022p-234\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-6-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E1\u003C\/sup\u003E\u003C\/span\u003E The number of social contacts at home arguably increases when other activities are restricted, but these contacts would likely be with the same individuals and would not constitute independent trials that could result in infections, as in a typical SEIR model. We therefore assume the contacts in the \u003Ckbd\u003Ehome\u003C\/kbd\u003E activity are unchanged, but our model could easily accommodate other assumptions.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-7\u0022\u003E\u003Cp id=\u0022p-235\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-7-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E2\u003C\/sup\u003E\u003C\/span\u003E Similarly to \u003Ca id=\u0022xref-ref-1-3\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-1\u0022\u003EAcemoglu et al. (2020)\u003C\/a\u003E, our approach is to characterize the frontier between deaths and economic losses. For any cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E, minimizing the objective function for that \u003Cem\u003E\u03c7\u003C\/em\u003E will yield a particular point on the frontier, and one can trace the frontier by varying \u003Cem\u003E\u03c7\u003C\/em\u003E. We focus on characterizing the frontier, rather than selecting an optimal point along the frontier. Selecting an optimum would entail determining a choice for the value of life, which is not desirable, as there is disagreement about the right value.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-8\u0022\u003E\u003Cp id=\u0022p-236\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-8-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E3\u003C\/sup\u003E\u003C\/span\u003E In our social contacts model, \u003Cem\u003Ec\u003C\/em\u003E\u003Csub\u003E\u003Cem\u003Eg,h\u003C\/em\u003E\u003C\/sub\u003E(\u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eg\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E), \u003Cstrong\u003E\u003Cem\u003El\u003C\/em\u003E\u003C\/strong\u003E\u003Csub\u003E\u003Cem\u003Eh\u003C\/em\u003E\u003C\/sub\u003E(\u003Cem\u003Et\u003C\/em\u003E)) is a polynomial.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-9\u0022\u003E\u003Cp id=\u0022p-237\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-9-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E4\u003C\/sup\u003E\u003C\/span\u003E We note that the ICU capacity in \u00cele-de-France started at 1,200 beds (\u003Ca id=\u0022xref-ref-25-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-25\u0022\u003EGodeluck 2020\u003C\/a\u003E) in the beginning of the pandemic, and was enhanced during the crisis, with a reported peak ICU utilization during the spring of 2020 of 2,668 beds, on April 8 2020 (\u003Ca id=\u0022xref-ref-45-1\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-45\u0022\u003ESterl\u00e9 et al. 2020\u003C\/a\u003E).\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-10\u0022\u003E\u003Cp id=\u0022p-238\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-10-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E5\u003C\/sup\u003E\u003C\/span\u003E We quantify the cost of death \u003Cem\u003E\u03c7\u003C\/em\u003E as a multiple of the annual GDP per capita in France, and use the shorthand notation \u003Cem\u003En\u00d7\u003C\/em\u003E to denote a value of \u003Cem\u003En\u003C\/em\u003E times this annual GDP per capita. For the GDP per capita of France, we use the figure for 2019, converting US dollars to euros using the exchange rate on June 17, 2020.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-11\u0022\u003E\u003Cp id=\u0022p-239\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-11-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E6\u003C\/sup\u003E\u003C\/span\u003E That each group should spend less time in confinement can be seen as a reasonable fairness requirement, e.g., consistent with Rawlsian justice (\u003Ca id=\u0022xref-ref-50-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-50\u0022\u003EYoung 1994\u003C\/a\u003E). In this sense, allowing ROLD increased levels of explicit discrimination has the potential to improve both efficiency and fairness.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-12\u0022\u003E\u003Cp id=\u0022p-240\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-12-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E7\u003C\/sup\u003E\u003C\/span\u003E This is due to the fact that a small fraction of the members of the 10-19 year old group are already in workforce. We do not count the value of lost schooling for them.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-13\u0022\u003E\u003Cp id=\u0022p-241\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-13-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E8\u003C\/sup\u003E\u003C\/span\u003E Note that we are counting the entire value for the 60-69 y.o. age group; this is due to the fact that, as we explain in \u003Ca id=\u0022xref-sec-47-2\u0022 class=\u0022xref-sec\u0022 href=\u0022#sec-47\u0022\u003ESection EC.4\u003C\/a\u003E, this value has already incorporated that only a fraction of the population in the 60-69 y.o. age group have work-eligible ages (60-64 y.o.), while the rest are retired.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-14\u0022\u003E\u003Cp id=\u0022p-242\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-14-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E9\u003C\/sup\u003E\u003C\/span\u003E In the expressions in (EC.27), (EC.29), (EC.33), we follow the convention that a product of matrices over an empty set of indices results in the identity matrix.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-15\u0022\u003E\u003Cp id=\u0022p-243\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-15-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E10\u003C\/sup\u003E\u003C\/span\u003E We retrieve the parameter values as reported before \u003Ca id=\u0022xref-ref-44-10\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-44\u0022\u003ESalje et al. (2020\u003C\/a\u003E) updated them on July 8, 2020.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-16\u0022\u003E\u003Cp id=\u0022p-244\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-16-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E11\u003C\/sup\u003E\u003C\/span\u003E We note that \u003Ca id=\u0022xref-ref-16-2\u0022 class=\u0022xref-bibr\u0022 href=\u0022#ref-16\u0022\u003EDu et al. (2020\u003C\/a\u003E) estimate the serial interval, and not the infectious period, to be 3.96 days. We borrow their confidence interval for the serial interval estimation and use it as an uncertainty range for our infectious period, which is of about the same length as their estimated serial interval.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-17\u0022\u003E\u003Cp id=\u0022p-245\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-17-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E12\u003C\/sup\u003E\u003C\/span\u003E This is a normalization to account for employees doing part time work.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-18\u0022\u003E\u003Cp id=\u0022p-246\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-18-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E13\u003C\/sup\u003E\u003C\/span\u003E For reference, we provide here the French rule. For a geographic \u201cdepartment\u201d in France to be declared under maximum alert, three criteria need to be satisfied. First criterion: the incidence rate, i.e. the number of positive cases per 100,000 inhabitants over seven days, must be greater than 250 per 100,000 inhabitants. Second criterion: the incidence rate among those over 65, the most vulnerable population, must exceed 100 cases per 100,000 inhabitants. Third criterion: more than 30% of resuscitation beds must be occupied by patients with COVID-19.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-19\u0022\u003E\u003Cp id=\u0022p-247\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-19-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E14\u003C\/sup\u003E\u003C\/span\u003E For the incidence rate, we count new cases using the Infectious state (and not the Exposed state) of the SEIR model; this is aligned with the assumption that a positive case can be detected through a viral test if and only if the individual is in the Infectious state.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-20\u0022\u003E\u003Cp id=\u0022p-248\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-20-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E15\u003C\/sup\u003E\u003C\/span\u003E Our targeted features include age group and activity themselves. To handle the categorical nature of activities, we create dummy indicator variables for each \u003Cem\u003Ea\u003C\/em\u003E.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-21\u0022\u003E\u003Cp id=\u0022p-249\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-21-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E16\u003C\/sup\u003E\u003C\/span\u003E For simplicity, we fold the lost wages term \u003Cspan class=\u0022inline-formula\u0022 id=\u0022inline-formula-83\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-83.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/inline-graphic-83.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\u003C\/span\u003E from our original model into \u003Cem\u003E\u03c7\u003C\/em\u003E.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-22\u0022\u003E\u003Cp id=\u0022p-250\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-22-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E17\u003C\/sup\u003E\u003C\/span\u003E To see this, fix all the problem parameters and define\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-117\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-150.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-150.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nThen \u003Cem\u003Eg\u003C\/em\u003E(\u003Cem\u003ET\u003C\/em\u003E) = 0, while\n\u003Cspan class=\u0022disp-formula\u0022 id=\u0022disp-formula-118\u0022\u003E\u003Cspan class=\u0022highwire-responsive-lazyload\u0022\u003E\u003Cimg src=\u0022data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\u0022 class=\u0022highwire-embed lazyload\u0022 alt=\u0022Embedded Image\u0022 data-src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-151.gif\u0022\/\u003E\u003Cnoscript\u003E\u003Cimg class=\u0022highwire-embed\u0022 alt=\u0022Embedded Image\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/highwire\/medrxiv\/early\/2021\/12\/15\/2021.03.23.21254155\/embed\/graphic-151.gif\u0022\/\u003E\u003C\/noscript\u003E\u003C\/span\u003E\n\n\u003C\/span\u003E\nbecause \u003Cem\u003Ee\u003C\/em\u003E\u003Csup\u003E\u003Cem\u003Ex\u003C\/em\u003E(\u003Cem\u003Ea\u003C\/em\u003E+\u003Cem\u003Eb\u003C\/em\u003E)\u003C\/sup\u003E \u003Cem\u003E\u2212 e\u003C\/em\u003E\u003Csup\u003E\u003Cem\u003Ex\u003C\/em\u003E(\u003Cem\u003E\u2212a\u003C\/em\u003E+\u003Cem\u003Eb\u003C\/em\u003E)\u003C\/sup\u003E \u003Cem\u003E\u2264\u003C\/em\u003E 0 for \u003Cem\u003Ea \u2265\u003C\/em\u003E 0, \u003Cem\u003Ex \u2264\u003C\/em\u003E 0. Therefore \u003Cem\u003Eg\u003C\/em\u003E is non-increasing in \u003Cem\u003Et\u003C\/em\u003E in [0, \u003Cem\u003ET\u003C\/em\u003E], and \u003Cem\u003Eg\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) \u003Cem\u003E\u2265\u003C\/em\u003E 0 for 0 \u003Cem\u003E\u2264 t \u2264 T\u003C\/em\u003E.\u003C\/p\u003E\u003C\/li\u003E\u003Cli class=\u0022fn\u0022 id=\u0022fn-23\u0022\u003E\u003Cp id=\u0022p-251\u0022\u003E\u003Ca class=\u0022rev-xref\u0022 href=\u0022#xref-fn-23-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cspan class=\u0022fn-label\u0022\u003E\u003Csup\u003E18\u003C\/sup\u003E\u003C\/span\u003E Since we use the simplification \u003Cem\u003ES\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003EE\u003C\/em\u003E (\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003EI\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) + \u003Cem\u003ER\u003C\/em\u003E(\u003Cem\u003Et\u003C\/em\u003E) \u003Cem\u003E\u2248 N\u003C\/em\u003E, it stands to reason that a per capita value is sufficient here.\u003C\/p\u003E\u003C\/li\u003E\u003C\/ul\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section ref-list\u0022 id=\u0022ref-list-1\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EReferences\u003C\/h2\u003E\u003Col class=\u0022cit-list ref-use-labels\u0022\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-1-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-1\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.1\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAcemoglu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EChernozhukov\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EV\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWerning\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EI\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWhinston\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EMD\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EOptimal targeted lockdowns in a multi-group SIR model\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EWorking Paper 27102, NBER\u003C\/span\u003E, URL \u003Ca href=\u0022http:\/\/dx.doi.org\/10.3386\/w27102\u0022\u003Ehttp:\/\/dx.doi.org\/10.3386\/w27102\u003C\/a\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-2-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-2\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.2\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAguilera\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJ\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ESome supermarkets are launching senior-only hours during the coronavirus pandemic\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003ENot all retailers think that\u2019s a good idea\u003C\/span\u003E. URL \u003Ca href=\u0022https:\/\/time.com\/5804574\/grocery-store-chains-seniors-coronavirus\/\u0022\u003Ehttps:\/\/time.com\/5804574\/grocery-store-chains-seniors-coronavirus\/\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJanuary 12, 2021\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-3-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-3\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.3\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAhn\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EHS\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESilberholz\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESong\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EX\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EX\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EOptimal covid-19 containment strategies: Evidence across multiple mathematical models\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EAvailable at SSRN 3834668\u003C\/abbr\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-4-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-4\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.4\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAlvarez\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EFE\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EArgente\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELippi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EF\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EA simple planning problem for COVID-19 lockdown\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EWorking Paper 26981, National Bureau of Economic Research\u003C\/span\u003E, URL \u003Ca href=\u0022http:\/\/dx.doi.org\/10.3386\/w26981\u0022\u003Ehttp:\/\/dx.doi.org\/10.3386\/w26981\u003C\/a\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-5-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-5\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-book\u0022 id=\u0022cit-2021.03.23.21254155v4.5\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAnderson\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ERM\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMay\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ERM\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E1992\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EInfectious Diseases of Humans: Dynamics and Control\u003C\/span\u003E (\u003Cspan class=\u0022cit-publ-name\u0022\u003EOxford University Press\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-6-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-6\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.6\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBastani\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EH\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDrakopoulos\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EK\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EGupta\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EV\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EVlachogiannis\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EI\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EHadjicristodoulou\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EC\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELagiou\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMagiorkinis\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EG\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EParaskevis\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ETsiodras\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EEfficient and targeted COVID-19 border testing via reinforcement learning\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003ENature\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E599\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E7883\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E108\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E113\u003C\/span\u003E, URL \u003Ca href=\u0022http:\/\/dx.doi.org\/10.1038\/s41586-021-04014-z\u0022\u003Ehttp:\/\/dx.doi.org\/10.1038\/s41586-021-04014-z\u003C\/a\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DNature%26rft.volume%253D599%26rft.spage%253D108%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-7-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-7\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.7\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBemporad\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2006\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EModel predictive control design: New trends and tools\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EProceedings of the 45th IEEE Conference on Decision and Control\u003C\/abbr\u003E, \u003Cspan class=\u0022cit-fpage\u0022\u003E6678\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E6683\u003C\/span\u003E (IEEE).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-8-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-8\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.8\u0022 data-doi=\u002210.1371\/journal.pone.0131080\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EB\u00e9raud\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EG\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2015\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EThe French connection: The first large population-based contact survey in France relevant for the spread of infectious diseases\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EPLOS ONE\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E10\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E7\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E1\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E22\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DPLOS%2BONE%26rft.volume%253D10%26rft.spage%253D1%26rft_id%253Dinfo%253Adoi%252F10.1371%252Fjournal.pone.0131080%26rft_id%253Dinfo%253Apmid%252F25919488%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003Ca href=\u0022\/lookup\/external-ref?access_num=10.1371\/journal.pone.0131080\u0026amp;link_type=DOI\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-doi cit-ref-sprinkles-crossref\u0022\u003E\u003Cspan\u003ECrossRef\u003C\/span\u003E\u003C\/a\u003E\u003Ca href=\u0022\/lookup\/external-ref?access_num=25919488\u0026amp;link_type=MED\u0026amp;atom=%2Fmedrxiv%2Fearly%2F2021%2F12%2F15%2F2021.03.23.21254155.atom\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-medline\u0022\u003E\u003Cspan\u003EPubMed\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-9-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-9\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.9\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBertsimas\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EIvanhoe\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJK\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EJacquillat\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EML\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPreviero\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELami\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EOS\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBouardi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EHT\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EOptimizing vaccine allocation to combat the covid-19 pandemic\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EmedRxiv\u003C\/abbr\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-10-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-10\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.10\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBirge\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJR\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ECandogan\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EO\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EFeng\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EY\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EReducing economic losses with targeted closures\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EWorking Paper 2020-57, University of Chicago, Becker Friedman Institute for Economics\u003C\/abbr\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-11-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-11\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.11\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBose\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESouyris\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMukherjee\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EUK\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EIvanov\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESeshadri\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EIII\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EACE\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EXu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EY\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EControl of epidemic spreads via testing and lock-down\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003ETechnical report\u003C\/abbr\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-12-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-12\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-book\u0022 id=\u0022cit-2021.03.23.21254155v4.12\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ECamacho\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EEF\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAlba\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ECB\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2013\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EModel predictive control\u003C\/span\u003E (\u003Cspan class=\u0022cit-publ-name\u0022\u003ESpringer science \u0026amp; business media\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-13-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-13\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.13\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EChang\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPierson\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EE\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EKoh\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EPW\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EGerardin\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ERedbird\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EB\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EGrusky\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELeskovec\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJ\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EMobility network models of COVID-19 explain inequities and inform reopening\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003ENature\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E589\u003C\/span\u003E:\u003Cspan class=\u0022cit-fpage\u0022\u003E1\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E6\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DNature%26rft.volume%253D589%26rft.spage%253D1%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-14-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-14\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.14\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ECot\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EC\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ECacciapaglia\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EG\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESannino\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EF\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EMining Google and Apple mobility data: Temporal anatomy for COVID-19 social distancing\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EScientific reports\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E11\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E1\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E1\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E8\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DScientific%2Breports%26rft.volume%253D11%26rft.spage%253D1%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-15-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-15\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.15\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDi Domenico\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPullano\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EG\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESabbatini\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ECE\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EB\u00f6elle\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EPY\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EColizza\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EV\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EImpact of lockdown on COVID-19 epidemic in \u00cele-de-France and possible exit strategies\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EBMC medicine\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E18\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E1\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E1\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E13\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DBMC%2Bmedicine%26rft.volume%253D18%26rft.spage%253D1%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-16-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-16\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.16\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDu Z\u003C\/span\u003E,  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EXu X\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EY\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWang\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ECowling\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EBJ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMeyers\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ESerial interval of COVID-19 among publicly reported confirmed cases\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EEmerging Infectious Diseases\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E26\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E6\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-17-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-17\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.17\u0022 data-doi=\u002210.1073\/pnas.2009033117\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDuque\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMorton\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EDP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESingh\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EB\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EZ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPasco\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ER\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMeyers\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ELA\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ETiming social distancing to avert unmanageable COVID-19 hospital surges\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EProceedings of the National Academy of Sciences\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E117\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E33\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E19873\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E19878\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DProceedings%2Bof%2Bthe%2BNational%2BAcademy%2Bof%2BSciences%26rft_id%253Dinfo%253Adoi%252F10.1073%252Fpnas.2009033117%26rft_id%253Dinfo%253Apmid%252F32727898%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003Ca href=\u0022\/lookup\/ijlink\/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoicG5hcyI7czo1OiJyZXNpZCI7czoxMjoiMTE3LzMzLzE5ODczIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjEvMTIvMTUvMjAyMS4wMy4yMy4yMTI1NDE1NS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-ijlink\u0022\u003E\u003Cspan\u003E\u003Cspan class=\u0022cit-reflinks-abstract\u0022\u003EAbstract\u003C\/span\u003E\u003Cspan class=\u0022cit-sep cit-reflinks-variant-name-sep\u0022\u003E\/\u003C\/span\u003E\u003Cspan class=\u0022cit-reflinks-full-text\u0022\u003E\u003Cspan class=\u0022free-full-text\u0022\u003EFREE \u003C\/span\u003EFull Text\u003C\/span\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-18-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-18\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.18\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDutta\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ER\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EGomes\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ESN\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EKalise\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPacchiardi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EUsing mobility data in the design of optimal lockdown strategies for the COVID-19 pandemic\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EPLoS Computational Biology\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E17\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E8\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003Ee1009236\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DPLoS%2BComputational%2BBiology%26rft.volume%253D17%26rft.spage%253D1009236e%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-19-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-19\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.19\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EEl Housni\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EO\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESumida\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EM\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ERusmevichientong\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ETopaloglu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EH\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EZiya\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECan testing ease social distancing measures? future evolution of covid-19 in nyc\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EarXiv preprint\u003C\/abbr\u003E\u003Cspan class=\u0022cit-pub-id-sep cit-pub-id-arxiv-sep\u0022\u003E \u003C\/span\u003E\u003Cspan class=\u0022cit-pub-id cit-pub-id-arxiv\u0022\u003E\u003Cspan class=\u0022cit-pub-id-scheme-arxiv\u0022\u003Earxiv:\u003C\/span\u003E2005.14700\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-20-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-20\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.20\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EEvgeniou\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ET\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EFekom\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EM\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EOvchinnikov\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPorcher\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ER\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPouchol\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EC\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EVayatis\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EN\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EEpidemic models for personalised covid-19 isolation and exit policies using clinical risk predictions\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EmedRxiv\u003C\/abbr\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-21-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-21\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.21\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EFavero\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ECA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EIchino\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ERustichini\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ERestarting the economy while saving lives under COVID-19\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EC.E.P.R Discussion Paper 14664, CEPR\u003C\/abbr\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-22-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-22\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.22\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EFoy\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EH\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ERussian businesses prepare for fresh lockdowns as Covid-19 cases soar\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EThe Financial Times\u003C\/span\u003E URL \u003Ca href=\u0022https:\/\/www.ft.com\/content\/f0696f19-fe0b-425c-8e73-3703bd44dec3\u0022\u003Ehttps:\/\/www.ft.com\/content\/f0696f19-fe0b-425c-8e73-3703bd44dec3\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJanuary 12, 2021\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-23-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-23\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.23\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EFrench Government\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EDonn\u00e9es hospitali\u00e8res relatives \u00e0 l\u2019\u00e9pid\u00e9mie de COVID-19\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.data.gouv.fr\/fr\/datasets\/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19\/\u0022\u003Ehttps:\/\/www.data.gouv.fr\/fr\/datasets\/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19\/\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EOctober 21, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-24-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-24\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.24\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EGiordano\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EG\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EColaneri\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EM\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDi Filippo\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBlanchini\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EF\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBolzern\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDe Nicolao\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EG\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESacchi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EColaneri\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBruno\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ER\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EModeling vaccination rollouts, SARS-CoV-2 variants and the requirement for non-pharmaceutical interventions in Italy\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003ENature Medicine\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E27\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E6\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E993\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E998\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DNature%2BMedicine%26rft.volume%253D27%26rft.spage%253D993%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-25-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-25\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.25\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EGodeluck\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECoronavirus : plus de 2.000 patients en r\u00e9animation en \u00cele-de-France\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003ELes Echos\u003C\/span\u003E, URL \u003Ca href=\u0022https:\/\/www.lesechos.fr\/economie-france\/social\/coronavirus-plus-de-2000-patients-en-reanimation-en-ile-de-france-1191003\u0022\u003Ehttps:\/\/www.lesechos.fr\/economie-france\/social\/coronavirus-plus-de-2000-patients-en-reanimation-en-ile-de-france-1191003\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 30, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-26-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-26\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.26\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EGoldstein\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJR\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ECassidy\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ET\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWachter\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EKW\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EVaccinating the oldest against COVID-19 saves both the most lives and most years of life\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EProceedings of the National Academy of Sciences\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E118\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E11\u003C\/span\u003E), ISSN \u003Cspan class=\u0022cit-issn\u0022\u003E0027-8424\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-27-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-27\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.27\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EGoogle\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003ECOVID-19 community mobility report\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.google.com\/covid19\/mobility\/\u0022\u003Ehttps:\/\/www.google.com\/covid19\/mobility\/\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EOctober 21, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-28-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-28\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.28\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EHarrison\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECoronavirus: Ireland\u2019s restrictions eased for over 70s\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EBBC\u003C\/span\u003E URL \u003Ca href=\u0022https:\/\/www.bbc.com\/news\/uk-northern-ireland-52547169\u0022\u003Ehttps:\/\/www.bbc.com\/news\/uk-northern-ireland-52547169\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJanuary 12, 2021\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-29-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-29\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.29\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EIlin\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EC\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAnnan-Phan\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ETai\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EXH\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMehra\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EHsiang\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBlumenstock\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJE\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EPublic mobility data enables COVID-19 forecasting and management at local and global scales\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EScientific reports\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E11\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E1\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E1\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E11\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DScientific%2Breports%26rft.volume%253D11%26rft.spage%253D1%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-30-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-30\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.30\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EKanbur\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EN\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAnkg\u00fcl\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EQuaranteenagers: A single country pandemic curfew targeting adolescents in Turkey\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EJournal of Adolescent Health\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E67\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E2\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-31-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-31\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.31\u0022 data-doi=\u002210.1016\/S1473-3099(20)30457-6\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EKucharski\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EAJ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EEffectiveness of isolation, testing, contact tracing, and physical distancing on reducing transmission of SARS-CoV-2 in different settings: a mathematical modelling study\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EThe Lancet Infectious Diseases\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E20\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E10\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E1151\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E1160\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DThe%2BLancet%2BInfectious%2BDiseases%26rft.volume%253D20%26rft.spage%253D1151%26rft_id%253Dinfo%253Adoi%252F10.1016%252FS1473-3099%252820%252930457-6%26rft_id%253Dinfo%253Apmid%252F32559451%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003Ca href=\u0022\/lookup\/external-ref?access_num=10.1016\/S1473-3099(20)30457-6\u0026amp;link_type=DOI\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-doi cit-ref-sprinkles-crossref\u0022\u003E\u003Cspan\u003ECrossRef\u003C\/span\u003E\u003C\/a\u003E\u003Ca href=\u0022\/lookup\/external-ref?access_num=32559451\u0026amp;link_type=MED\u0026amp;atom=%2Fmedrxiv%2Fearly%2F2021%2F12%2F15%2F2021.03.23.21254155.atom\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-medline\u0022\u003E\u003Cspan\u003EPubMed\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-32-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-32\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.32\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELehot\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EM\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBorgne\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EBL\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECovid-19 : ces chiffres qui montrent que Paris a d\u00e9pass\u00e9 le seuil d\u2019alerte maximale depuis le 25 septembre\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EFranceinfo\u003C\/abbr\u003E accessed October 5, 2020.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-33-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-33\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.33\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EML\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBouardi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EHT\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELami\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EOS\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ETrikalinos\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ETA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ETrichakis\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ENK\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBertsimas\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EForecasting covid-19 and analyzing the effect of government interventions\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EMedRxiv\u003C\/abbr\u003E \u003Cspan class=\u0022cit-fpage\u0022\u003E2020\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E06\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-34-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-34\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.34\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMagid\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJ\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECabinet approves removal of age restriction for those returning to work\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EThe Times of Israel\u003C\/span\u003E URL \u003Ca href=\u0022https:\/\/www.timesofisrael.com\/liveblog-may-2-2020\/\u0022\u003Ehttps:\/\/www.timesofisrael.com\/liveblog-may-2-2020\/\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJanuary 12, 2021\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-35-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-35\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-book\u0022 id=\u0022cit-2021.03.23.21254155v4.35\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMas-Colell\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWhinston\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EMD\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EGreen\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJR\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E1995\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EMicroeconomic theory\u003C\/span\u003E, volume \u003Cspan class=\u0022cit-vol\u0022\u003E1\u003C\/span\u003E (\u003Cspan class=\u0022cit-publ-name\u0022\u003EOxford University Press\u003C\/span\u003E \u003Cspan class=\u0022cit-publ-loc\u0022\u003ENew York\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-36-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-36\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.36\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMatrajt\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EEaton\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELeung\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ET\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBrown\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EER\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EVaccine optimization for COVID-19\u003C\/span\u003E, \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003Ewho to vaccinate first?\u003C\/abbr\u003E\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-37-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-37\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.37\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMorris\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EDH\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ERossine\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EFW\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPlotkin\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJB\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELevin\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ESA\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EOptimal, near-optimal, and robust epidemic control\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003ECommunications Physics\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E4\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E1\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E1\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E8\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DCommunications%2BPhysics%26rft.volume%253D4%26rft.spage%253D1%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-38-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-38\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.38\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPataro\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EIML\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EA control framework to optimize public health policies in the course of the COVID-19 pandemic\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EScientific Reports\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E11\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E1\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E13403\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DScientific%2BReports%26rft.volume%253D11%26rft.spage%253D13403%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-39-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-39\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.39\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPerakis\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EG\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESinghvi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESkali Lami\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EO\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EThayaparan\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECOVID-19: A multipeak SIR based model for learning waves and optimizing testing\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EAvailable at SSRN 3817680\u003C\/abbr\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-40-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-40\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.40\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPrem\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EK\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ECook\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EAR\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EJit\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EM\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2017\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EProjecting social contact matrices in 152 countries using contact surveys and demographic data\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EPLoS computational biology\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E13\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E9\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003Ee1005697\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DPLoS%2Bcomputational%2Bbiology%26rft.volume%253D13%26rft.spage%253De1005697%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-41-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-41\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.41\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPrem\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EK\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EThe effect of control strategies to reduce social mixing on outcomes of the COVID-19 epidemic in Wuhan, China: a modelling study\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EThe Lancet Public Health\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E5\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E5\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003Ee261\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003Ee270\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DThe%2BLancet%2BPublic%2BHealth%26rft.volume%253D5%26rft.spage%253D261e%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-42-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-42\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.42\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EReuters\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2021\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EFrance goes under nationwide 6pm curfew as Covid-19 death toll surpasses 70,000\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EFrance24\u003C\/span\u003E URL \u003Ca href=\u0022https:\/\/www.france24.com\/en\/europe\/20210116-france-set-for-nationwide-6pm-curfew-in-effort-to-stem-covid-19\u0022\u003Ehttps:\/\/www.france24.com\/en\/europe\/20210116-france-set-for-nationwide-6pm-curfew-in-effort-to-stem-covid-19\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJanuary 12, 2021\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-43-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-43\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.43\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EReuters Staff\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EBosnian region eases lockdown on seniors, children after court ruling\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EReuters\u003C\/span\u003E URL \u003Ca href=\u0022https:\/\/www.reuters.com\/article\/us-health-coronavirus-bosnia-idUSKCN2261RA\u0022\u003Ehttps:\/\/www.reuters.com\/article\/us-health-coronavirus-bosnia-idUSKCN2261RA\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJanuary 12, 2021\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-44-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-44\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.44\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESalje\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EH\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EEstimating the burden of SARS-CoV-2 in France\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EScience 38\u003C\/abbr\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-45-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-45\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.45\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESterl\u00e9\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EC\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EAlexandre\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EV\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003Ede Livonni\u00e8re\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECoronavirus en \u00eele-de-France : 13 000 malades encore hospitalis\u00e9s, d\u00e9j\u00e0 40 000 tests positifs\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003ELe Parisien\u003C\/span\u003E, URL \u003Ca href=\u0022https:\/\/www.leparisien.fr\/paris-75\/coronavirus-en-ile-de-france-13-000-malades-encore-hospitalises-deja-40-000-tests-positifs-29-04-2020-8307801.php\u0022\u003Ehttps:\/\/www.leparisien.fr\/paris-75\/coronavirus-en-ile-de-france-13-000-malades-encore-hospitalises-deja-40-000-tests-positifs-29-04-2020-8307801.php\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 30, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-46-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-46\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.46\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ETiirinki\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EH\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECOVID-19 pandemic in Finland\u2013preliminary analysis on health system response and economic conse-quences\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EHealth policy and technology\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E9\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E4\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E649\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E662\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DHealth%2Bpolicy%2Band%2Btechnology%26rft.volume%253D9%26rft.spage%253D649%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.47\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWellenius\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EGA\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EImpacts of state-level policies on social distancing in the United States using aggregated mobility data during the COVID-19 pandemic\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EarXiv e-prints\u003C\/abbr\u003E arXiv\u20132004.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-48-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-48\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.48\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWille\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EVan Hoang\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ET\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EFunk\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EColetti\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBeutels\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EHens\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EN\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ESOCRATES: an online tool leveraging a social contact data sharing initiative to assess mitigation strategies for COVID-19\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EBMC Research Notes\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E13\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E293\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-49-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-49\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.49\u0022 data-doi=\u002210.1073\/pnas.2010836117\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EXiong\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EC\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EHu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ES\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EYang\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EM\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELuo\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EW\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EZhang\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EMobile device data reveal the dynamics in a positive relationship between human mobility and COVID-19 infections\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EProceedings of the National Academy of Sciences\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E117\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E44\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E27087\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E27089\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DProceedings%2Bof%2Bthe%2BNational%2BAcademy%2Bof%2BSciences%26rft_id%253Dinfo%253Adoi%252F10.1073%252Fpnas.2010836117%26rft_id%253Dinfo%253Apmid%252F33060300%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003Ca href=\u0022\/lookup\/ijlink\/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoicG5hcyI7czo1OiJyZXNpZCI7czoxMjoiMTE3LzQ0LzI3MDg3IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjEvMTIvMTUvMjAyMS4wMy4yMy4yMTI1NDE1NS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-ijlink\u0022\u003E\u003Cspan\u003E\u003Cspan class=\u0022cit-reflinks-abstract\u0022\u003EAbstract\u003C\/span\u003E\u003Cspan class=\u0022cit-sep cit-reflinks-variant-name-sep\u0022\u003E\/\u003C\/span\u003E\u003Cspan class=\u0022cit-reflinks-full-text\u0022\u003E\u003Cspan class=\u0022free-full-text\u0022\u003EFREE \u003C\/span\u003EFull Text\u003C\/span\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-50-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-50\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-book\u0022 id=\u0022cit-2021.03.23.21254155v4.50\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EYoung\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EHP\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E1994\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EEquity\u003C\/span\u003E (\u003Cspan class=\u0022cit-publ-name\u0022\u003EPrinceton University Press\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.51\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E \u003Cspan class=\u0022cit-name-given-names\u0022\u003EYuan\u003C\/span\u003E \u003Cspan class=\u0022cit-name-surname\u0022\u003EYx\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2015\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ERecent advances in trust region algorithms\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EMathematical Programming\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E171\u003C\/span\u003E:\u003Cspan class=\u0022cit-fpage\u0022\u003E249\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E281\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DMathematical%2BProgramming%26rft.volume%253D171%26rft.spage%253D249%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003C\/ol\u003E\u003C\/div\u003E\u003Cdiv class=\u0022section ref-list\u0022 id=\u0022ref-list-2\u0022\u003E\u003Ch2 class=\u0022\u0022\u003EE-companion References\u003C\/h2\u003E\u003Col class=\u0022cit-list ref-use-labels\u0022\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-52-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-52\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.52\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EAgence R\u00e9gionale de Sant\u00e9 \u2014 \u00cele-de-France\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003ED\u00e9confinement: strat\u00e9gie de d\u00e9pistage et recherche de cas-contacts\u003C\/span\u003E. URL \u003Ca href=\u0022https:\/\/www.iledefrance.ars.sante.fr\/deconfinement-strategie-de-depistage-et-recherche-de-cas-contacts\u0022\u003Ehttps:\/\/www.iledefrance.ars.sante.fr\/deconfinement-strategie-de-depistage-et-recherche-de-cas-contacts\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 30, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-53-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-53\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.53\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EBanque de France\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020a\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EPoint sur la conjoncture fran\u00e7aise \u00e0 fin avril 2020\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.banque-france.fr\/sites\/default\/files\/media\/2020\/06\/10\/point-conjoncture_avril-2020.pdf\u0022\u003Ehttps:\/\/www.banque-france.fr\/sites\/default\/files\/media\/2020\/06\/10\/point-conjoncture_avril-2020.pdf\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 19, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-web no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.54\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EBanque de France\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020b\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EPoint sur la conjoncture fran\u00e7aise \u00e0 fin mai 2020\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.banque-france.fr\/sites\/default\/files\/media\/2020\/06\/11\/point-conjoncture-09_juin-2020-20200609.pdf\u0022\u003Ehttps:\/\/www.banque-france.fr\/sites\/default\/files\/media\/2020\/06\/11\/point-conjoncture-09_juin-2020-20200609.pdf\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 19, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.55\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBi\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EQ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EEpidemiology and transmission of COVID-19 in 391 cases and 1286 of their close contacts in Shenzhen, China: a retrospective cohort study\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EThe Lancet Infectious Diseases\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E20\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E8\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-56-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-56\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.56\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDiekmann\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EO\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EHeesterbeek\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EJAP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ERoberts\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EMG\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2010\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EThe construction of next-generation matrices for compartmental epidemic models\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EJournal of The Royal Society Interface\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E7\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E47\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E873\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E885\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DJournal%2Bof%2BThe%2BRoyal%2BSociety%2BInterface%26rft.volume%253D7%26rft.spage%253D873%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.57\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDu Z\u003C\/span\u003E,  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EXu X\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EY\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EWang\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ECowling\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EBJ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMeyers\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EL\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ESerial interval of COVID-19 among publicly reported confirmed cases\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EEmerging Infectious Diseases\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E26\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E6\u003C\/span\u003E).\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.58\u0022 data-doi=\u002210.1073\/pnas.2009033117\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDuque\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ED\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMorton\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EDP\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESingh\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EB\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EDu\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EZ\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPasco\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ER\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMeyers\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003ELA\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ETiming social distancing to avert unmanageable COVID-19 hospital surges\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EProceedings of the National Academy of Sciences\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E117\u003C\/span\u003E(\u003Cspan class=\u0022cit-issue\u0022\u003E33\u003C\/span\u003E):\u003Cspan class=\u0022cit-fpage\u0022\u003E19873\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E19878\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DProceedings%2Bof%2Bthe%2BNational%2BAcademy%2Bof%2BSciences%26rft_id%253Dinfo%253Adoi%252F10.1073%252Fpnas.2009033117%26rft_id%253Dinfo%253Apmid%252F32727898%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003Ca href=\u0022\/lookup\/ijlink\/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoicG5hcyI7czo1OiJyZXNpZCI7czoxMjoiMTE3LzMzLzE5ODczIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjEvMTIvMTUvMjAyMS4wMy4yMy4yMTI1NDE1NS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-ijlink\u0022\u003E\u003Cspan\u003E\u003Cspan class=\u0022cit-reflinks-abstract\u0022\u003EAbstract\u003C\/span\u003E\u003Cspan class=\u0022cit-sep cit-reflinks-variant-name-sep\u0022\u003E\/\u003C\/span\u003E\u003Cspan class=\u0022cit-reflinks-full-text\u0022\u003E\u003Cspan class=\u0022free-full-text\u0022\u003EFREE \u003C\/span\u003EFull Text\u003C\/span\u003E\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.59\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EFrench Government\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EDonn\u00e9es hospitali\u00e8res relatives \u00e0 l\u2019\u00e9pid\u00e9mie de COVID-19\u003C\/abbr\u003E. \u003Ca href=\u0022https:\/\/www.data.gouv.fr\/fr\/datasets\/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19\/\u0022\u003Ehttps:\/\/www.data.gouv.fr\/fr\/datasets\/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19\/\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EOctober 21, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-web no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.60\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EGoogle\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003ECOVID-19 community mobility report\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.google.com\/covid19\/mobility\/\u0022\u003Ehttps:\/\/www.google.com\/covid19\/mobility\/\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EOctober 21, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-61-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-61\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.61\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EINSEE\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2016a\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003ET402: Salaire brut en \u00e9quivalent temps plein, par secteur d\u2019activit\u00e9, r\u00e9gion et d\u00e9partement\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.insee.fr\/fr\/statistiques\/fichier\/4204500\/T402.xls\u0022\u003Ehttps:\/\/www.insee.fr\/fr\/statistiques\/fichier\/4204500\/T402.xls\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 19, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-62-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-62\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.62\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EINSEE\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2016b\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003ET404 : Salaire brut en\u00e9quivalent temps plein, par tranche d\u2019\u00e2ge simplifi\u00e9e, cat\u00e9gorie socioprofessionnelle simplifi\u00e9e et r\u00e9gion\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.insee.fr\/fr\/statistiques\/fichier\/4204500\/T404.xls\u0022\u003Ehttps:\/\/www.insee.fr\/fr\/statistiques\/fichier\/4204500\/T404.xls\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 19, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-63-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-63\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.63\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EINSEE\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2019\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EAu quatrieme trimestre 2019, letaux de ch\u00f4mage passe de 8,5 % a 8,1 %\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.insee.fr\/fr\/statistiques\/4309346\u0022\u003Ehttps:\/\/www.insee.fr\/fr\/statistiques\/4309346\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 19, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-64-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-64\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.64\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth cit-collab\u0022\u003EINSEE\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003EEstimation de population par d\u00e9partement, sexe et\u00e2ge quinquennal - ann\u00e9es 1975 \u00e0 2020\u003C\/span\u003E. \u003Ca href=\u0022https:\/\/www.insee.fr\/fr\/statistiques\/fichier\/1893198\/estim-pop-dep-sexe-aq-1975-2020.xls\u0022\u003Ehttps:\/\/www.insee.fr\/fr\/statistiques\/fichier\/1893198\/estim-pop-dep-sexe-aq-1975-2020.xls\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJune 19, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-web no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.65\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ELehot\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EM\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EBorgne\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EBL\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECovid-19 : ces chiffres qui montrent que Paris a d\u00e9pass\u00e9 le seuil d\u2019alerte maximale depuis le 25 septembre\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EFranceinfo\u003C\/span\u003E accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EOctober 5, 2020\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-66-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-66\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.66\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EMohammad\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EH\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003ECoronavirus : un habitant de Bobigny consid\u00e9r\u00e9 comme le nou-veau \u201cpatient z\u00e9ro\u201d\u003C\/span\u003E. \u003Cspan class=\u0022cit-source\u0022\u003EFrance Bleu\u003C\/span\u003E, URL \u003Ca href=\u0022https:\/\/www.francebleu.fr\/infos\/sante-sciences\/coronavirus-un-habitant-de-bobigny-considere-comme-le-nouveau-patient-zero-1588796619\u0022\u003Ehttps:\/\/www.francebleu.fr\/infos\/sante-sciences\/coronavirus-un-habitant-de-bobigny-considere-comme-le-nouveau-patient-zero-1588796619\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJanuary 9, 2021\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-67-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-67\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal\u0022 id=\u0022cit-2021.03.23.21254155v4.67\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003EPerasso\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EA\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2018\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EAn introduction to the basic reproduction number in mathematical epidemiology\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EESAIM: Proceedings and Surveys\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E62\u003C\/span\u003E:\u003Cspan class=\u0022cit-fpage\u0022\u003E123\u003C\/span\u003E\u2013\u003Cspan class=\u0022cit-lpage\u0022\u003E138\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003Ca href=\u0022{openurl}?query=rft.jtitle%253DESAIM%253A%2BProceedings%2Band%2BSurveys%26rft.volume%253D62%26rft.spage%253D123%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx\u0022 class=\u0022cit-ref-sprinkles cit-ref-sprinkles-openurl cit-ref-sprinkles-open-url\u0022\u003E\u003Cspan\u003EOpenUrl\u003C\/span\u003E\u003C\/a\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Cdiv class=\u0022cit ref-cit ref-journal no-rev-xref\u0022 id=\u0022cit-2021.03.23.21254155v4.68\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESalje\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EH\u003C\/span\u003E\u003C\/span\u003E, \u003Cspan class=\u0022cit-etal\u0022\u003Eet al.\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-article-title\u0022\u003EEstimating the burden of SARS-CoV-2 in France\u003C\/span\u003E. \u003Cabbr class=\u0022cit-jnl-abbrev\u0022\u003EScience\u003C\/abbr\u003E \u003Cspan class=\u0022cit-vol\u0022\u003E38\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003Cli\u003E\u003Cspan class=\u0022ref-label ref-label-empty\u0022\u003E\u003C\/span\u003E\u003Ca class=\u0022rev-xref-ref\u0022 href=\u0022#xref-ref-69-1\u0022 title=\u0022View reference  in text\u0022 id=\u0022ref-69\u0022\u003E\u21b5\u003C\/a\u003E\u003Cdiv class=\u0022cit ref-cit ref-web\u0022 id=\u0022cit-2021.03.23.21254155v4.69\u0022\u003E\u003Cdiv class=\u0022cit-metadata\u0022\u003E\u003Ccite\u003E\u003Cspan class=\u0022cit-auth\u0022\u003E\u003Cspan class=\u0022cit-name-surname\u0022\u003ESterl\u00e9\u003C\/span\u003E  \u003Cspan class=\u0022cit-name-given-names\u0022\u003EC\u003C\/span\u003E\u003C\/span\u003E (\u003Cspan class=\u0022cit-pub-date\u0022\u003E2020\u003C\/span\u003E) \u003Cspan class=\u0022cit-source\u0022\u003E\u00cele-de-France: y aura-t-il assez de tests Covid avant N\u00f6el?\u003C\/span\u003E \u003Ca href=\u0022https:\/\/www.leparisien.fr\/societe\/sante\/ile-de-france-y-aura-t-il-assez-de-tests-covid-avant-noel-18-12-2020-8414952.php\u0022\u003Ehttps:\/\/www.leparisien.fr\/societe\/sante\/ile-de-france-y-aura-t-il-assez-de-tests-covid-avant-noel-18-12-2020-8414952.php\u003C\/a\u003E, accessed \u003Cspan class=\u0022cit-date-in-citation\u0022\u003EJanuary 28, 2021\u003C\/span\u003E.\u003C\/cite\u003E\u003C\/div\u003E\u003Cdiv class=\u0022cit-extra\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/li\u003E\u003C\/ol\u003E\u003C\/div\u003E\u003Cspan class=\u0022highwire-journal-article-marker-end\u0022\u003E\u003C\/span\u003E\u003C\/div\u003E\u003Cspan class=\u0022related-urls\u0022\u003E\u003C\/span\u003E\u003C\/div\u003E\u003C\/div\u003E  \u003C\/div\u003E\n\n  \n  \u003C\/div\u003E\n\u003C\/div\u003E\n  \u003C\/div\u003E\n\u003C\/div\u003E\n\u003C\/div\u003E\u003Cscript type=\u0022text\/javascript\u0022 src=\u0022https:\/\/www.medrxiv.org\/sites\/default\/files\/js\/js_zP7WWIfzbyzvaM63L39cNV2juU_1XVH7wduFK9gcMNI.js\u0022\u003E\u003C\/script\u003E\n\u003C\/body\u003E\u003C\/html\u003E"}