COVID-19 ON Response Framework Model

Simulation and model-fitting code for a model that describes SARS-CoV-2 transmission in Ontario, Canada.

The purpose of this project was to extend the work done in this paper (https://doi.org/10.1101/2021.03.26.21254421) by implementing the government of Ontario's colour coded response framework as a method of mitigating the effects of COVID-19 in the province.

Previous iterations of this model include:

• https://github.com/k3fair/COVID-19-ON-model (Manuscript: https://doi.org/10.1101/2021.03.26.21254421) and
• https://github.com/VadimKar/COVID-19-Hierarchy-model (Manuscript: https://doi.org/10.1073/pnas.2014385117)

To run a set of simulations, make all required option selections (detailed within each script) and launch script.

All data needed to run scripts are included in the "InputFiles" folder, with the exception of the parameter set inputs for covidHierV48rf_Github_basesim.R, as the large number of files makes this infeasible. However, such parameter sets can be generated using covidHierV48rf_Github_parmfit.R. Expected runtime varies between scripts, but an individual simulation should run within 1 minute.

Scripts

• multiplot.R contains a function that combines multiple plots into one image. A dependency for the scripts below.

• covidHierV48rf_Github_parmfit.R runs fitting procedure for obtaining parameter sets, outputs files containing the best parameter set as determined by that run of the fitting algorithms.

• covidHierV48rf_Github_basesim.R evaluates output from covidHierV46_2Github_parmfit.R to identify parameter sets resulting in the best fit to data, runs simulations using those parameter sets, and outputs files containing the results of that identification and simulation.

• covidHierV48rf_Github_counterfactuals.R runs simulations for counterfactuals, outputs results of these simulations. Can be used to reproduce results shown in Figure 1, 2, 9, and 10.

Input files

• cmat_data_weighted.rds contains contact matrices; data adapted from https://github.com/kieshaprem/synthetic-contact-matrices

• covidHierData.rds contains several objects. Some are artefacts of a previous version of the model (https://github.com/VadimKar/COVID-19-Hierarchy-model) retained within the file to preseve the dimensions of certain data structures - these are denoted with an asterisk. Objects are as follows:

  • Distd50*: the proportion of people voluntarily distancing by day 21 of the outbreak (data from https://news.gallup.com/opinion/gallup/298310/americans-step-social-distancing-even-further.aspx)

  • testRat: estimate of the ratio of true infections to positive tests (from https://doi.org/10.1101/2020.03.14.20036178)

  • Msave: Travel matrix, where entries are numbers of commuters between a home and workplace county. Data from Statistics canada, 2016 census, catalogue no. 98-400-x2016391.

  • storeSatIncTrm.65*: Pre-calculated values of baseline transmission probability (across values of c and xi) which leads to 65% of population being infected after 1 year without mitigation

  • caseCtBin*: the daily number of reported cases, with smaller counties grouped by population density (data from https://www.publichealthontario.ca/en/data-and-analysis/infectious-disease/covid-19-data-surveillance/covid-19-data-tool)

  • testT3.1*: function giving the (smoothed) testing intensity on any given day t from the day the 50th case positive case was reported (i.e., t=0 if fewer than 50 cumulative cases reported). These values increase from 0 to 1following the (smoothed) relative increase in number of tests taken each day based on data from https://www.ontario.ca/page/2019-novel-coronavirussection-0

  • CountyPopDensities*: population densities in each county (data from https://www12.statcan.gc.ca/health-sante/82-228/details/page.cfm?Lang=E&Tab=1&Geo1=HR&Code1=3570&Geo2=PR&Code2=35&Data=Rate&SearchText=York%20Regional%20Health%20Unit&SearchType=Contains&SearchPR=01&B1=All&Custom=&B2=All&B3=All)

• mobilitydat-REAL_to2021-02-27_v4.rds contains data on mobility trends for retail and recreational locations (adapted from https://www.google.com/covid19/mobility/).

• ONdat_casesbyPHU_to2021-08-10.rds contains data on new reported COVID-19 cases in Ontario, grouped by PHU (data from https://covid-19.ontario.ca/data).

• ONdat_newKage_to2021-08-10.rds contains data on new reported COVID-19 cases in Ontario, grouped by age-class (data from https://covid-19.ontario.ca/data).

• PHUtoREGIONlinker_numeric.rds contains a key for linking Ontario's census divisions (regions) to the Public Health Units (PHU) they belong to.

• response_framework_refined_2021-05-01.rds contains data on the zone tiers that each PHU in Ontario was in during it's implementation (data from https://data.ontario.ca/dataset/ontario-covid-19-zones)

• simsFITPARMS_v48rf.rds contains best-fit parameter sets (generated from covidHierV48rf_Github_parmfit.R and covidHierV48rf_Github_basesim.R), used for all counterfactual simulations

• tests_by_PHU_aug112021.rds contains testing data by PHU in Ontario (data from https://data.ontario.ca/dataset/ontario-covid-19-testing-metrics-by-public-health-unit-phu)

System & hardware requirements

• Windows 10 Pro Version 2004

• Sufficient RAM to support storage of data structures during simulations

Installation guides

All software should install within a few minutes on a standard computer, the versions listed here are those the scripts have been tested on.

• R Version 4.0.3 https://www.r-project.org/

• R Studio Version 1.4.1717 (IDE for R) https://rstudio.com/