Modeling Human Temporal and Spatial Structured Contacts for Epidemic Prediction

Human travel and social contacts are deemed as the key force in driving the transmission of infectious diseases. Consequently, epidemic models should represent individuals' spatiotemporal contacts in detail to investigate the complex process of epidemic diffusion. Mathematical epidemic models usually assume a homogeneous human population and describe unstructured contacts. Most network-based epidemic models do not synthetically consider the temporal and spatial features of human structured contact behaviors. We here combined a social network with a bipartite network to build a multilayer agent network to model heterogeneous individuals' temporal and spatial structured contacts. We used the largest collective outbreak of H1N1 influenza at a Chinese university in 2009 as a case study. Experimental results indicate that our models can reproduce individuals' daily travel and social contact patterns, as well as the H1N1 influenza outbreak. We found that only quarantining dormitories to stop interbuilding transmission could not achieve a great effect in mitigating epidemic outbreaks at a university. The prohibition of students' visiting across dormitory rooms was indispensable to prevent intrabuilding transmission of infectious diseases. Furthermore, it would be better to quarantine admitted case patients' close contacts to control potentially latent individuals.