The dynamics underlying global spread of emerging infectious diseases

Abstract

BACKGROUND: Recent outbreaks of emerging infectious diseases (EIDs), including SARS, pandemic influenza, MERS-CoV, Ebola, and Zika virus, have caused substantial health and economic burden. Large-scale computational models parameterized with worldwide air network (WAN) and populations have been the mainstream tool for studying global spread of EIDs since the 1980s. In addition to advanced global epidemic simulators such as GLEAM, recent analytical studies have partially revealed how epidemic arrival time (EAT) for different populations in the WAN depends on epidemic parameters and the network features of the WAN. Our objective is to explicitly characterize the dynamics underlying global spread of EIDs. METHOD: We developed a novel probabilistic framework based on nonhomogeneous Poisson Process (NPP) to characterize global spread of EIDs. Specifically, our framework entails (i) modelling exportations of cases from the epidemic origin as NPPs; (ii) accounting for the effect of high outgoing air traffic (the ‘hub-effect’) and continuous seeding on local epidemic growth rate and mobility rate; (iii) modeling the effect of multiple paths using superposition of NPPs. To verify the accuracy of our framework, we developed a stochastic global epidemic simulator comprising more than 3,000 airports and 30,000 flight connections. RESULTS: Comparing the simulated EAT with the analytically derived EAT, we showed that our analytical framework can provide very good estimates of EAT for almost all populations in the WAN. CONCLUSION: We reveal that the EATs in WAN-based global meta-population models can be analytically measured with high accuracy from the epidemiologic and network parameters. In pursuit for analytical insights, we explicitly characterize how the dynamics of global spread of EIDs depend on the underlying epidemiologic and network properties.