Data-driven modelling of spreading processes

Models of contagion processes conventionally rely on unstructured populations and the homogenous mixing assumption, while the importance of networks have been shown recently to be crucial to determine their critical behaviour and final outcome. Earlier studies built on synthetic networks provided the advantage to be analytically treatable but were hardly applicable for real world scenarios. However, in the last ten years the access to high resolution human behavioural datasets from mobile devices, communication, and pervasive technologies has propelled a wealth of developments in the analysis of social networks and human mobility. Such datasets can be integrated into models of spreading processes leading to hybrid data-driven models where data is engaged with synthetic spreading dynamics for more realistic models of contagion phenomena.
In this talk we will take an overview on the different ways of data-driven modelling of spreading processes, like information spreading, social or biological contagion. We will discuss how and what kind of data can be useful, how they can be engaged with modelled dynamical processes, and what knowledge we can gain with these techniques. We will pay special attention on data-driven modelling of large-scale epidemics and we will discuss the latest efforts taken to understand the COVID-19 pandemic using data-driven models of epidemic processes.