Multivariate spatial Poisson mixtures with applications in disease source classification.

摘要

Gastrointestinal disease (GID) data obtained from the Canadian Institute for Health Information (CIHI) provides motivation to extend Mixture Model (MM) literature in order to classify disease based on infection source. Specifically, MMs are employed to classify GID data as foodborne or waterborne. This work will account for spatially indexed disease using two methods. In the first case, independent conditionally autoregressive spatial priors will be assigned to the log linear term of each of the mixture components (one per disease source). The second case investigates a non-independence assumption. These two models (identified as the IMCAR and MCAR models) are contrasted with the standard CAR spatial model described by Besag et al, 1991. All models are compared via a simulation study, with application to Alberta GID data (1992-1998).