Spatial disease mapping using the Poisson-Gamma model

Abstract

In disease mapping, it is preferable to estimate the risk rather than the significance in general, but the variation in estimation precision across the geographical map of the study region must also be taken into consideration. In such a situation the conventional methods would not yield the best estimates. Heterogeneity is an important aspect to be considered as significant in Disease Mapping and relative risk estimation. The simple regression models often do not capture the extent of the variation exhibited in the spatial count data. This is the case when the spatial data is over-dispersed or there is spatial correlation due to unobserved confounders. In such situations, it is appropriate to include some additional terms, which may be in the form of the prior distribution. In this paper, a Poisson model with Gamma prior is used to model and map the dengue incidences in Tamil Nadu to explain the patterns of variations.