City-Scale Simulation Model of COVID-19 Spread

Forecasting models for COVID-19 at the national and state levels have been common, but few have accounted for SARS-CoV-2 transmission at city or smaller scales. In this talk, we describe two complementary Mathematica epidemiological simulation models—one an agent-based model (ABM), the other a compartmental model based on system dynamics (SD)—which have been developed to forecast COVID-19 spread at city scale and below. The COVID ABM simulates the movements, contacts, behaviors and disease transmission outcomes of every person in a city, installation or facility as they go about their daily activity schedules. To facilitate the COVID ABM, a discrete-event simulation framework specific to agent-based models was developed that is generally applicable to ABM applications. The SD COVID model represents subunits of the population distinguished by disease state as COVID-19 spreads through the population. Mathematica is found to easily and elegantly express essential concepts of agent-based, epidemiological and system dynamics modeling as well as discrete event simulation. The talk concludes on how agent-based modeling and system dynamics can work together to best inform public health decision making.