Predicting spatial transmission at the early stage of epidemics on a networked metapopulation

Realization of accurate real-time predictions of infectious diseases is an important but challenging task, because spatial transmission among networked populations is stochastic and time-varying. In this paper, we propose a new algorithm to predict the susceptible subpopulations which will be infected in the next time step at the early stage of an epidemic on a metapopulation network by using data of infection and topology. We first estimate the epidemic infection rate, which helps us to infer the increment of newly infected individuals during a unit time. Then we predict the possible infected subpopulations by ranking the infected likelihoods of corresponding susceptible subpopulations. The simulation results on the Barabasi-Albert scale-free metapopulation network verify the performance of our algorithm.