Methods for High-Fidelity Epidemic Simulations on Time-Varying Networks

Abstract

While there are numerous models for studying the spread of infectious diseases, many researchers rely on fully-mixed models, which assume random-mixing of individuals. While these models are straightforward, they hardly mirror what we observe in the real-world. In practice, each person has a finite set of contacts with whom they interact. Network epidemic models explicitly model this set of contacts, allowing us to capture complex individual-level structure and behaviors. In this work, we demonstrate how network models can be used to deepen our understanding of infectious diseases and interventions designed to control their spread. Leveraging time-varying empirical networks, we first illustrate the merits of a network-based approach to epidemic modeling with specific emphasis on capturing individual heterogeneities and multifaceted epidemic control strategies. We then broaden our aperture, examining temporal resolution and spreading phenomena over time-varying networks. Technological advances will continue to pave the way for easier and more streamlined collection of high-fidelity network data. However, this begs the question: how much temporal granularity is needed to accurately capture spreading phenomena over networks? We explore the type and quality of data needed to study spreading processes with accuracy and precision. This research will provide public health practitioners with essential tools to enhance our understanding of infectious diseases and their interventions using realistic network-based approaches.