Simulation Modeling of Infectious Disease and the COVID-19 Epidemic

Mark S. Roberts, MD, MPP @UPittsburgh

Video Recording

Slides

Abstract:
This talk will introduce the use of agent-based simulation modeling as a methodology to understand the spread of an infectious disease and the potential impact of mitigation strategies. The talk will introduce FRED, the Framework for Reconstructing Epidemiologic Dynamics, an agent-based simulation modeling tool constructed over the past decade by the Public Health Dynamics Laboratory (PHDL) under multiple grants from NIGMS under the Modeling Infectious Disease Agents Study (MIDAS). The tool represents the entire population of the US through a synthetic population, which is statistically equivalent to the actual population in any geographic location. Individual agents in the model can develop diseases, pass the disease onto others based on contact networks such as neighborhoods, schools and workplaces. The tool has been used to evaluate the impact of strategies for the H1N1 epidemic, Avian influenza and measles, and has been used by policy makers to assist in legislative decisions.

We have incorporated a model of COVID-19 into FRED, and individuals proceed through their disease and may become ill enough to require hospitalization, be transferred to intensive care, and require mechanical ventilation. The model has been calibrated with data from multiple different sources, including the CDC and multiple state databases. The model produces epidemic curves that chart the number of hospitalizations, and the number of patients requiring ICUs and ventilators, as well as the number of deaths and the number of recovered individuals. We present the potential impact of multiple different mitigation strategies, such as social distancing and closing schools.

Bio:
Mark S. Roberts, MD, MPP is Professor and Chair of Health Policy and Management, and holds secondary appointments in Medicine, Industrial Engineering, and Clinical and Translational Science. A practicing general internist, he has conducted research in decision analysis and the mathematical modeling of disease for over 30 years, and has expertise in cost effectiveness analysis, mathematical optimization and simulation, and the measurement and inclusion of patient preferences into decision problems. He has used decision analysis to examine clinical, costs, policy and allocation questions in liver transplantation, vaccination strategies, operative interventions, and the use of many medications. His recent research has concentrated in the use of mathematical methods from operations research and management science, including Markov Decision Processes, Discrete Event and Agent-based Simulation.  As director of the Public Health Dynamics Laboratory, he continues to lead the development of simulation tools for representing complex diseases and the evaluation of policies to improve health and public health. 

His methodological interests in decision sciences, cost effectiveness analysis, comparative effectiveness, operations research, simulation modeling, clinical research methods, quality of life and utility analysis, and inference in observational studies; his content interests in health care financing and physician and patient incentives, transplantation, HIV care, diagnostic tests, the opioid epidemic, preventive care and tailoring clinical guidelines to individual patients.