Abstract or Additional Information
Heterogeneity in contact network structure is known to play an important role in the spread of epidemics. However, models tracking exact network structure quickly become intractable as the network grows in size. Moment closure techniques (e.g. pair approximations) have been widely used and, more recently, tractable edge-based models have been introduced on random networks with a specified degree distribution. A factor that further complicates disease dynamics is that contact patterns change in response to infection. We present a stochastic model for an epidemic on a dynamic multilayer network, where layers represent different modes of transmission. We derive the large graph limit of the stochastic process in order to obtain a model that retains key features but is amenable to analysis. In doing so, we are able to obtain some results on edge-based models and also characterize under what conditions a pair approximation gives the exact large graph limit. Based on the limit theorem, we formulate a hybrid stochastic-deterministic model that allows for efficient parameter estimation. The motivations for this framework in Ebola modeling will also be briefly discussed.