Dipartimento di Scienze Statistiche

Giovedì 12 marzo 2026 dalle ore 14.00    Aula 34 del Dipartimento di Scienze Statistiche   "Mean-field limit and parameter estimation for contagion dynamics on higher-order structures"   il Prof. Nicos Georgiou della University of Sussex, UK,    Classical models of contagion epidemics have graph based representations where the nodes represent the population and infections occur from an infected individual to a susceptible individual via a network link between them. At the same time, healing mechanisms try to fight the epidemic at node level. The simple descriptions of these representations also underscore their limitations, since the propagation of contagions in complex systems often involves repeated or simultaneous stimuli that individuals receive from their social contacts (such as their work or household). One may mechanistically encode these non-linear effects into contagion models by considering interactions that go beyond simple pairwise connections, by using hypergraphs rather than graphs. In this talk we will see the role higher-order interactions can play in the behaviour of the model when a rigorous mean-field can be obtained. Furthermore, we will discuss some inference techniques for the parameters of the epidemic (rates of infection) when only partial information on the type of infection is available and with only a single realisation of the epidemic curve as a datum.