Mycobacterium Tuberculosis Hotspots: A Spatial Analysis of Tuberculosis Cases in West Java, Indonesia

Abstract

Globally, tuberculosis ranks as the second most prevalent cause of mortality resulting from a single infectious disease. In 2023, Indonesia experienced a 24-year peak in tuberculosis cases, with West Java Province exhibiting the highest occurrence rate, reflecting a 24.2% increase compared to the previous year. This study aims to investigate the determinants of tuberculosis cases in West Java by employing a spatial approach to account for regional heterogeneity. The Geographically Weighted Negative Binomial Regression (GWNBR) method is applied due to its suitability for handling count data with overdispersion while accommodating spatially varying relationships. Parameter estimation is conducted using the Maximum Likelihood Estimation (MLE) method enhanced with the Nelder--Mead optimization algorithm. Four kernel function types are incorporated in the modeling process, namely Fixed Gaussian, Adaptive Gaussian, Fixed Bi-square, and Adaptive Bi-square. Based on the Akaike Information Criterion (AIC), the Adaptive Gaussian kernel demonstrates the best model performance, achieving an AIC value of 1,000.11. Population density emerges as the most influential determinant, showing statistical significance across all regions. These findings emphasize the importance of developing localized tuberculosis control strategies that are tailored to specific regional characteristics.