Loading...

Accurate data-driven models for epidemic spreading are crucial to the understanding of epidemic dynamics and the design of suitable non-pharmaceutical interventions. One main feature of social networks is group mixing, as highlighted by the widely observed differences in the density of social ties between and within groups of similar individuals in real-world networks. In this talk, I will address graph models for heterogeneous networks with a planted partition and their application to characterize epidemic spreading at the urban scale.
I will present two random graph models — the Fitness-Corrected Block Model (FCBM) and the Hidden-degree Geometric Block Model (HGBM) — describing their main features, discussing their strengths and limitations, and suggesting possible extensions. I will then show how the FCBM can be used to define a geospatial social network based on publicly available data and empirical sociological findings. Under the assumption that age and geographic distance are the main drivers for physical interactions, I will present analytical results and simulations that help characterise the diffusion of the disease in the population and the surrounding territory.