covid19model-wpro

Application of the Imperial College covid19model to countries in the Western Pacific for the WHO Western Pacific Regional Office (WPRO) and to provide modelling support to countries in the region.

This repository is a fork of the original code used for modelling estimated deaths and infections for COVID-19 from "Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe", Flaxman, Mishra, Gandy et al, Nature, 2020, the published version of the original Report 13 published by MRC Centre for Global Infectious Disease Analysis, Imperial College London: Estimating the number of infections and the impact of nonpharmaceutical interventions on COVID-19 in 11 European countries

The code used is in the wpro_analysis branch (default) off the forked master branch of the Version 4 release. Updated upstream code (to Version 7 release) is stored in the master branch with unnecessary code for our analysis is stored in the imperial-code/ directory. Please go to the upstream repository to see the original code and for version details and instructions in the associated README.md file. Results for the original model applied to European countries are available here with the technical model description available here and here

Changes to upstream code

For the analysis we made the following key changes to the model and modelling code:

• We changed the serial interval fit to have a Weibull distribution with mean = 5.5 and sd = 2.9 in line with the generation time distribution reported in Ferretti et al.
• We changed the prior for mu to have a mean of 2.7 to be more in-line with meta-analyses of R0 estimates (for example see here).
• We created a separate script to fetch John Hopkins University data as an alternative to ECDC data. ECDC data seems to be lagged by a day for Western Pacific countries.
• We created a separate results and figures generation script wpro-results.r rather than using the base wpro.r script.

Countries

So far we are applying this code to the following countries:

• Philippines
• Malaysia

How to run the code

There are two ways to run the code:

• Open the rstudio project covid19model.Rproj file in rstudio and run/source wpro.r file
• To run from the command line please enter the cloned directory and type Rscript wpro.r base in terminal

Run mode settings

Different run modes incorporated into the wpro.r script: :

• include_ncd which specifies if the weighted-IFR estimate includes the relative differences in all-cause mortality between the specific country and where the IFR estimates are from (to account for co-morbidities and environmental factors potentially affecting COVID-19 deaths).
• npi_on to turn on the effect of NPIs.
• fullRun which must always be used if you want to obtain reliable results.
• debug quick run and extra outputs for debugging.
• model to specify the stan model used.
• useJHU set to use data from John Hopkins University rather than the ECDC. ECDC data seems to be a day late for Western Pacific countries.
• endDate to specify the end date of data to run the model on. If set to NULL then runs to the last date in the fetched data.

Results

To produce results we have written a separate script wpro-results.r which needs to be run/sourced after the main wpro.r script has been run. Note the JOBID number needs to be entered manually into the script to produce the results. We store results and figures for different runs by last date in the fetched data.

• The results are stored in two folders results and figures.
• Results have the stored stan fits and data used for plotting
• Figures have the images with daily cases, daily death and Rt for all countries.