Research Colloquium on Computational Social Science/Data Sciences

Covasim: an open-source agent-based model of COVID-19 dynamics and interventions by Author Cliff Kerr, Dina Mistry, Robyn Stuart, Jamie Cohen, Romesh Abeysuriya, Daniel Klein on behalf of the IDM COVID Response Team 

The COVID-19 pandemic has created an urgent need for models that can project epidemic trends, explore intervention scenarios, and estimate resource needs. Here we describe the methodology and applications of Covasim (COVID-19 Agent-based Simulator), an open-source model developed to help address these questions. Covasim aims to capture the nuances of realistic COVID-19 transmission, including detailed representations of demographics and transmission network structure (including schools and aged care facilities), intrahost viral dynamics, and an extensive list of pharmaceutical and non-pharmaceutical interventions. Covasim has been designed to be adaptable to different contexts and accessible to different users, with simple Python installation, extensive documentation and usage examples, software unit and regression tests and an interactive webapp. To date, Covasim has been used and extended by collaborators in nearly a dozen countries, including being used to inform policy decisions in the US, UK, Kenya, Australia, and Eswatini. We will highlight findings from two recent studies: (1) quantifying what coverage levels of testing, contact tracing, and mask usage are sufficient for epidemic control in the Seattle context; and (2) determining the epidemic impacts of school reopening, and how these can be mitigated through different classroom measures. 

Cliff Kerr is a Senior Research Scientist at the Institute for Disease Modeling, where he has been leading the development of the Covasim model. Prior to COVID, his work focused on optimization algorithm development, modeling of family planning interventions, and value-of-information analyses. He has a B.Sc. in neuroscience from the University of Queensland, a Ph.D. in theoretical physics from the University of Sydney, and a Diploma of Arts in composition from the Sydney Conservatorium of Music. His postdoctoral work, as part of a DARPA project at the SUNY Downstate Medical Center, looked at how detailed computer simulations of a monkey brain could be used to teach robotic arms to pick up balls. With funding from the Australian Research Council, he extended this work in his own lab at the University of Sydney to explore the neural basis of computation, determining that this problem is very hard. Simultaneously, he was a co-founder and lead software developer for the Optima Consortium for Decision Science, a nonprofit consultancy that has helped the governments of more than 50 countries plan investments for HIV, tuberculosis, malaria, and childhood nutrition using the Optima suite of open-source modeling tools.