We have created a geospatially explicit agent-based model (ABM) to simulate hospital capacity for North Carolina (NC) hospitals during the coronavirus disease 2019 (COVID-19) pandemic. This ABM is used to predict capacity strain on both ICU and non-ICU beds for both COVID-19 and non-COVID-19 patients. The public version of this model in this repository is not fully calibrated to actual data because of data restrictions. Specifically, access to counts of COVID-19 and non-COVID-19 patients by hospital is restricted. Without access to the actual data, this repository is not intended to be used to reproduce actual estimates but instead intended to illustrate the theoretical components of the model.
- This code is released with no support.
- This repository is copied from an internal repository and the commit history is removed, so the development history is not included in the public repository.
Due to the time it takes to run the model, we recommend running production runs using parallel processing on multiple CPUs. Follow the instructions in scripts/README.md before launching a production run.
A production run requires a local installation of python to create the necessary files. Using your preferred virtual environment mechanism (virtualenv, conda, etc), create the environment first by install the requirements:
pip install -r docker/requirements.txtThe command used to launch the production run from root while in the virtualenv is:
bash ./scripts/run_all.sh
The Overview, Design Concepts, Details (ODD) Protocol document is available on arXiv. This document provides complete details on how the model works.
This page serves as an introduction to the project. Please see this README.md for model details.