Speaker:
Fan Bu, PhD
Abstract: Understanding the dynamics of infectious disease transmissions requires models that account for both transmission processes and the evolving contact patterns underlying them. We propose a generative framework that couples a stochastic epidemic model with a dynamic, adaptive social network structure, allowing infection rates to vary with individual-level covariates and network evolution to depend on disease status. This joint process is formulated as a continuous-time Markov chain, capturing the interplay between disease transmission and contact network changes. To address the challenge of inference from partially observed epidemic data, we develop data-augmentation algorithms that efficiently impute missing infection and recovery times while respecting the constraints imposed by the dynamic network structure. Such data augmentation strategy can be implemented through a fully Bayesian Markov chain Monte Carlo (MCMC) scheme or a stochastic expectation-maximization (EM) algorithm. Through extensive simulations and applications to real-world influenza transmission data with high-resolution contact tracing, we demonstrate the efficiency and accuracy of our inference approaches in recovering epidemic parameters with uncertainty quantification. Our framework provides a new tool for integrating emerging multi-modality epidemiological data to study important heterogeneities in disease transmission.
Bio: Fan Bu is an Assistant Professor in Biostatistics at the University of Michigan. She received her PhD in Statistics from Duke University and was a postdoctoral research fellow at UCLA Department of Biostatistics. Fan's research focuses on developing Bayesian (sometimes non-Bayesian) and computational methods for time-varying, networked, or distributed data, with applications in infectious disease modeling, health data science, and computational social science.
Location: Hock Plaza Auditorium
Zoom Link: https://duke.zoom.us/j/98778470668