Covasim is a stochastic agent-based simulator designed to be used for COVID-19 (novel coronavirus, SARS-CoV-2) epidemic analyses. These include projections of indicators such as numbers of infections and peak hospital demand. Covasim can also be used to explore the potential impact of different interventions, including social distancing, school closures, testing, contact tracing, and quarantine.
The scientific paper describing Covasim is available at http://paper.covasim.org. The recommended citation is:
Kerr CC, Stuart RM, Mistry D, Abeysuriya RG, Hart G, Rosenfeld R, Selvaraj P, Núñez RC, Hagedorn B, George L, Izzo A, Palmer A, Delport D, Bennette C, Wagner B, Chang S, Cohen JA, Panovska-Griffiths J, Jastrzębski M, Oron AP, Wenger E, Famulare M, Klein DJ (2020). Covasim: an agent-based model of COVID-19 dynamics and interventions. Institute for Disease Modeling. Available at http://paper.covasim.org.
The Covasim webapp is available at http://app.covasim.org.
Questions or comments can be directed to [email protected], or on this project's GitHub page. Full information about Covasim is provided in the documentation.
Contents
Python >=3.6 (64-bit). (Note: Python 2 is not supported.)
We also recommend, but do not require, using Python virtual environments. For more information, see documentation for venv or Anaconda.
Install with pip install covasim
. If everything is working, the following Python commands should bring up a plot:
import covasim as cv sim = cv.Sim() sim.run() sim.plot()
Clone a copy of the repository. If you intend to make changes to the code, we recommend that you fork it first.
(Optional) Create and activate a virtual environment.
Navigate to the root of the repository and install the Covasim Python package using one of the following options:
To install with web app support (recommended):
python setup.py develop
To install as a standalone Python model without webapp support:
python setup.py develop nowebapp
To install Covasim and optional dependencies (be aware this may fail since it relies on private packages), enter:
python setup.py develop full
The module should then be importable via
import covasim
.
There are several examples in the examples directory. These can be run as follows:
python examples/simple.py
This example creates a figure using default parameter values.
python examples/run_sim.py
This shows a slightly more detailed example, including creating an intervention and saving to disk.
python examples/run_scenarios.py
This shows a more complex example, including running an intervention scenario, plotting uncertainty, and performing a health systems analysis.
All core model code is located in the covasim
subfolder; standard usage is
import covasim as cv
. The other subfolders, cruise_ship
, data
, and webapp
, are
also described below.
The model consists of two core classes: the Person
class (which contains
information on health state), and the Sim
class (which contains methods for
running, calculating results, plotting, etc.).
The structure of the covasim
folder is as follows:
base.py
: TheParsObj
class, plus basic methods of theBaseSim
class, and associated functions.defaults.py
: The default colors, plots, etc. used by Covasim.interventions.py
: TheIntervention
class, for adding interventions and dynamically modifying parameters.misc.py
: Miscellaneous helper functions.parameters.py
: Functions for creating the parameters dictionary and loading the input data.population.py
: ThePeople
class, for handling updates of state for each person.plotting.py
: Plotly graphs to supplement the built-in Matplotlib graphs.population.py
: Functions for creating populations of people, including age, contacts, etc.requirements.py
: A simple module to check that imports succeeded, and turn off features if they didn't.run.py
: Functions for running simulations (e.g. parallel runs and theScenarios
class).sim.py
: TheSim
class, which performs most of the heavy lifting: initializing the model, running, and plotting.utils.py
: Functions for choosing random numbers, many based on Numba, plus other helper functions.version.py
: Version, date, and license information.
A version of the Covasim model specifically adapted for modeling the Diamond
Princess cruise ship. It uses its own parameters file (parameters.py
) and has
slight variations to the model (model.py
).
This folder contains loading scripts for the epidemiological data in the root data
folder, as well as data on age distributions for different countries and household sizes.
For running the interactive web application. See the webapp README for more information.
Please see the readme in each subfolder for more information.
This folder contains a command-line interface (CLI) version of Covasim; example usage:
covasim --pars "{pop_size:20000, pop_infected:1, n_days:360, rand_seed:1}"
Note: the CLI is currently not compatible with Windows. You will need to add this folder to your path to run from other folders.
Scripts to automatically scrape data (including demographics and COVID epidemiology data), and the data files themselves (which are not part of the repository).
This folder contains the Dockerfile
and other files that allow Covasim to be
run as a webapp via Docker.
This folder contains demonstrations of simple Covasim usage.
Licensing information and legal notices.
Integration, development, and unit tests.
Utilities for hyperparameter sweeps, using Weights and Biases. See the sweep README for more information.
The code in this repository was developed by IDM to support our research in disease transmission and managing epidemics. We’ve made it publicly available under the Creative Commons Attribution-Noncommercial-ShareAlike 4.0 License to provide others with a better understanding of our research and an opportunity to build upon it for their own work. We make no representations that the code works as intended or that we will provide support, address issues that are found, or accept pull requests. You are welcome to create your own fork and modify the code to suit your own modeling needs as contemplated under the Creative Commons Attribution-Noncommercial-ShareAlike 4.0 License. See the contributing and code of conduct READMEs for more information.