Zhilan Feng
$500,000
James W Moody
Charles L Nunn
Duke University
North Carolina
Mathematical and Physical Sciences (MPS)
In today's highly connected world, the prevention, prediction, and control of epidemics is of paramount importance for global health, economic productivity, and geopolitical stability. Numerous infectious disease outbreaks over the past two decades have demonstrated the need for epidemiological modeling. They also revealed shortcomings of existing scientific techniques to accurately predict epidemic dynamics and to devise effective control strategies. This project will establish a new efficient simulation method that makes it possible to assess rare but highly consequential events. It will be used to identify decisive risk factors concerning the fabric of virus-spreading interactions that can facilitate large epidemic outbreaks. A well-documented example are superspreading events that played an important role in the COVID-19 pandemic. The investigations will be focused on models for diseases similar to COVID-19 and HIV as archetypal cases. The improved understanding and models of epidemiological processes will be used to devise and analyze efficient preventive strategies with the goal of providing more reliable guidance for the general public and health-policy decision makers, saving lives and resources.<br/><br/>Traditionally, the dynamics of infectious diseases are studied on the basis of deterministic compartmental models, where the population is divided into large groups, and deterministic differential equations for the group sizes are employed to investigate disease dynamics. Classical examples are the deterministic SIR and SIS models. This is a strong simplification of reality that ignores to a large extent the heterogeneity in contact patterns and biomedically relevant attributes across the population as well as the stochastic nature of infection processes. Both have a decisive impact on the dynamics at the early stages of epidemic outbreaks and need to be incorporated to enable reliable predictions. Markov-chain Monte Carlo methods can sample more realistic stochastic agent-based dynamics, but cannot efficiently assess the preconditions leading to rare consequential events. The project will address this challenge with a new numerical technique that allows one to efficiently sample important but rare epidemic trajectories of realistic models under suitable constraints. The research will renew attention on the crucial role of rare events in the genesis of large outbreaks, including combinations of bottlenecks in contact networks and the stochastic nature of the disease dynamics. Risk-factor analysis based on the new method will provide answers to cutting-edge questions in disease diffusion concerning outbreak preconditions, information flow, and control strategies. This approach will open new avenues for research on the prevention and control of epidemics.<br/><br/>This project is jointly funded by the Mathematical Biology program of the Division of Mathematical Sciences (DMS) in the Directorate for Mathematical and Physical Sciences (MPS) and the Human Networks and Data Science program (HNDS) of the Division of Behavioral and Cognitive Sciences (BCS) in the Directorate for Social, Behavioral and Economic Sciences (SBE).<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.