Endemic-epidemic models with discrete-time serial interval distributions for infectious disease prediction

Multivariate count time series models are an important tool for analyzing and predicting the spread of infectious disease. We consider endemic-epidemic framework, a class autoregressive disease surveillance counts, replace default autoregression on counts from previous period with more flexible weighting schemes inspired by discrete-time serial interval distributions. employ three different parametric formulations, each additional unknown parameter estimated via profile likelihood approach, compare them to unrestricted nonparametric approach. The new methods illustrated in univariate analysis dengue fever incidence San Juan, Puerto Rico, spatiotemporal study viral gastroenteritis 12 districts Berlin. assess predictive performance suggested several reference at various forecast horizons. In both applications, is considerably improved proposed schemes.