Measurability of the epidemic reproduction number in data-driven contact networks

Liu, Quan-Hui; Ajelli, Marco; Aleta, Alberto; Merler, Stefano; Moreno, Yamir; Vespignani, Alessandro

Briefly, the obtained synthetic population allows us to define a multilayer network of contacts that can be used to model the infection of the influenza virus in the population. Each layer represents one of the main four settings where influenza transmission occurs, namely households, schools, workplaces, and the general community [2, 3, 4] . The household layer is composed by n h disconnected components, each one representing one household (i.e., the total number of households is n h ). Each household is composed by a certain amount of nodes, each one representing one single individual. The amount of nodes in each household is determined by sampling from the actual Italian household size distribution. An age is associated to each node by sampling from the actual Italian age distribution by household size using an algorithm tailored to match realistic age-gaps between household members (details in Fumanelli et al. [1] ). Each node lives also in the community layer, which is composed by one single giant fully connected component. By sampling from the multinomial distribution of schooling and employment rates by age, for each node is determined whether it also lives in the school or workplace layer. Analogously to what has been described for the household layer, the workplace layer is composed by n w disconnected components, each one representing one workplace. The number of nodes in each workplace is sampled from the actual Italian workplace size distribution. The school layer is composed by n s disconnected components, each one representing one school. The number of nodes in each school is sampled from the actual Italian school size distribution by school level (e.g., pre-school, primary school, middle school, high school). Details in Fumanelli et al. [1] .