The Minnesota COVID-19 Model

© 2020 Regents of the University of Minnesota. This software is licensed via the GNU General Public License 3.0.

This source code is provided "as-is" free of charge for use and modification, with attribution. Use of the code requires knowledge of the R programming language and mathematical modeling.

Important code information:

• The provided R code is pre-programmed to run scenarios 1-4 in version 3.0 of the Minnesota COVID-19 model.
• Model parameters are set to base case values by default.
• To generate ICU demand outcomes as reported in our results, run scenario code with unconstrained ICU capacity. To do so uncomment line 92 of scenarios_script.R .
• To generate all other outcomes, run scenario code with default ICU capacity constraints.
• Provided code can also be used to generate plots of outcomes over time (ICU demand, prevalent infections, cumulative deaths, etc.).
• Using the code assumes basic knowledge of R and use of mathematical models. Technical support, over and above the existing documentation and FAQs (refer to links below), is unavailable.

For more information (including Scenario 1-4 results and model documentation): https://www.sph.umn.edu/research/projects/covid-19-model-code/

How to use:

First, navigate to MNCOVID19/R and open MNCOVID19.Rproj. This will automatically set your working directory and ensure that the correct files are accessed by the various scripts in the repository.

Next, open the setup_model.R file (MNCOVID19/R/setup_model.R) and ensure that all of the required packages are installed on your machine.

The primary script that is used to run the model and visualize the output is R/scenarios_script.R. Sourcing setup_model.R at the top of the script will source all of the necessary model functions and load the required packages. Running the rest of the code in scenarios_script.R will run the model, print summaries of the model outputs for each strategy, and generate plots for visualizing outputs.

Code orientation

Below is a general orientation to the primary .R files in this repository and the model functions they define:

• parameters_v3.R: Defines the function parameters(), which is used to define and calculate model input parameters. The output is a list of parameters, which is then passed to other model functions.
• covid_19_model_function_v3.R: Defines the main model function, covid_19_model_function(), which does some intermediate calculations on input parameters and calculates the change in compartments (i.e., dS, dE, dI, etc.) for one time step. Changes to the contact age-mixing matrix are applied within this function to reflect time-varying social distancing orders and policies.
• solve_model.R: Defines function solve_model(), which takes as input model parameters and other simulation inputs (e.g. time horizon) and calls the covid_19_model_function() at each time step to determine the change in compartment sizes. It aggregates model output at each time step into a single output matrix.

Other .R files include:

• convert_parms.R: Defines the function convert_parms() that adjusts the subset of model parameters that need to be adjusted for the chosen model timestep (parms$timestep).
• lambda.R: Defines the function calculate_lambda() that calculates the age-specific force of infection given the current contact matrix and prevalence of infection.
• process_output_v3.R: Defines the function process_output() that extracts and saves measures of interest from raw model outputs (e.g. prevalence, daily deaths, ICU demand, cumulative infections, cumulative deaths, etc.).