Speaker
Description
Disease transmission unfolds on contact networks that are inherently dynamic, with interactions appearing and disappearing over time. Recent advances in GPS- and survey-based data collection have enabled reconstruction of time-dependent contact networks for entire communities. However, these observations are incomplete and noisy, making the resulting networks uncertain representations of the true transmission process. Here, we perform a simulation study investigating how observational uncertainty in dynamic contact networks propagates into epidemiological forecasts and intervention assessments. To characterize network structure, we utilize betweenness centrality alongside a topological data analysis tool that extracts a multiscale summary of how mesoscopic connectivity patterns emerge, persist, and reorganize over time. Using these complementary summaries, we quantify how broad intervention strategies (e.g. vaccination), reshape topology of dynamic contact networks, and examine how those structural changes relate to epidemic outcomes in SIR agent-based models defined on the same networks. We further conduct computational experiments across multiple levels of observation variance, measuring resulting changes in network summaries alongside shifts in epidemic outcomes. Ultimately, we aim to apply this framework to GPS-derived contact networks from rural Madagascar to identify structural signatures of sensitivity to public health interventions in real-world dynamic networks.
