Spatial Scan Statistics for Models with Excess Zeros and Overdispersion

Abstract

Spatial Scan Statistics usually assume Poisson or Binomial distributed data, which is not realistic in many disease surveillance scenarios. We propose a statistical model for disease cluster detection, through a modification of the spatial scan statistic to account for inflated zeros and overdispersion simultaneously. A computer program is implemented using the Expectation-Maximization algorithm to solve the latent variables. Numerical simulations are shown to assess the effectiveness of the method. We present results for Hanseniasis surveillance in the Brazilian Amazon using this technique, compared with other models.