Identifying Influential Pandemic Regions Using Graph Signal Variation

Developing methods to analyse infection spread is an impor-tant step in the study of pandemic and containing them. Theprincipal mode for geographical spreading of pandemics isthe movement of population across regions. We are interestedin identifying regions (cities, states, or countries) which areinfluential in aggressively spreading the disease to neighbor-ing regions. We consider a meta-population network withSIR (Susceptible-Infected-Recovered) dynamics and developgraph signal-based metrics to identify influential regions.Specifically, a local variation and a temporal local variationmetric is proposed. Simulations indicate usefulness of the lo-cal variation metrics over the global graph-based processingsuch as filtering.