Application of Bayesian Spatial Method in Analyzing Tetanus Cases in Central Java Using Binomial-Poisson Model with R

This study investigates the spatial distribution of tetanus cases in Central Java, Indonesia, using a Bayesian spatial method with a Binomial-Poisson model. The analysis addresses the challenge of excess zeros in the data, common in rare diseases like tetanus, by incorporating a zero-inflated Poisson distribution. The Bayesian framework utilizes the Besag-York-Mollié (BYM2) model to account for spatial autocorrelation and employs complexity-penalized priors for the latent level process. The study leverages the Integrated Nested Laplace Approximation (INLA) for efficient and accurate inference. Results provide insights into the spatial patterns of tetanus cases in Central Java, potentially informing targeted interventions and public health strategies. The suggested Bayesian mixed models for Tetanus cases in 2022 across 36 districts/cities in Central Java, Indonesia, incorporating explanatory variables from various dimensions, including health, demographics, industry, and education. Significant dimensions with Tetanus cases are predominantly found in demographic and industrial aspects. Through the Deviance Information Criterion (DIC), the Binomial-Poisson mixture distribution is identified as the most suitable for the data, with an excess of zero values at approximately 62.8% and significant extreme observations in specific regions.