Contact networks across ages: computational analysis of epidemiological dynamics

For decades mathematical modeling in epidemiology has helped understanding the dynamics of infectious diseases, as well as describe possible intervention scenarios to prevent and control them. However, such models were relying on several assumptions, such as the ones on the structure of the underlying contact networks. The robustness of their predictions was therefore limited by this lack of knowledge. About a decade ago, with the advent of digital epidemiology, scientist have finally started to try to corroborate those assumptions, for instance with the use of wearable sensors to measure indeed contact networks. In this thesis, together with collaborators, I try to combine the digital collection of public health data with computational tools, in order to have a more realistic understanding of the phenomena under consideration. In two projects it was possible to finalise such marriage, thanks also to fruitful collaborations with other researchers who provided the data. This is for instance the case for the two chapters respectively on the modeling of influenza and plague outbreaks. Although they involve different technologies for the data collection, historical epochs and data types, the traditional epidemiological modeling allowed us to derive interpretable conclusions, capable for instance to inform public health interventions. In other projects, either the relevant data collection is still ongoing in the lab (like for the FoodRepo project), or the data collection has not started yet (like for the project on measles), although our work provides insights on the importance of such data collection for future studies. In the first chapter, we explore different mechanistic interpretation compatible with our data on the 1630 plague outbreak in Venice, collected through the digitisation of parish books from the historical State Archives of Venice. The data shows a non trivial temporal structure, which led us to propose few different epidemiological explanations. Further data collection will be needed to better constrain such interpretations. In the second chapter, we use previously recorded contact data in a high-school to assess the relative effect of ventilation on the influenza spread, with respect to vaccination strategies. Our result suggest the usefulness of non pharmacological interventions such as indeed improved ventilation, which become even more meaningful in the context of vaccination hesitancy and low vaccine efficacy, due for instance to the high mutation rate of viruses like influenza virus. In the third chapter, we propose a simple network generation model to try to explain differences in the incidence of highly infectious diseases (such as measles), across countries with similar vaccination coverages. Such differences are indeed one of the main open questions in public health, which are not yet fully understood even considering social phenomena such as recent anti-vax movements. In the last chapter, we present our open database of ba