Bayesian spatial-temporal autologistic regression model of dengue hemorrhagic fever cases in East Java, Indonesia

The purpose of this study is to discuss and develop Spatial-Temporal Autologistic Regression Model (STARM) to represent spreading of the Aedes aegypti which is indicated by the endemic level of DHF (Dengue Hemorrhagic Fever) in East Java. The method that be used to estimate STARM parameter was Bayesian method with Markov Chain Monte Carlo (MCMC) and Gibbs Sampler simulation. This study observed 38 districts as spatial lattice units, meanwhile temporal unit is represented by monthly period of evidence (January-December) in 2002-2008. Result of the research was obtained STARM model that indicate the spreading pattern of the Aedes aegypti that is indicated by the endemic level of DHF incidence in East Java have spatially and temporally positive correlation. Model validation using 95% confidence interval shows that all estimators are significant. This is also supported by a MAE value 0.09 and the percentage of correctly classified predicted data 90%, which means there are 90 correctly classified data of 100 prediction data.