Application of Geographically Weighted Panel Regression (GWPR) on Tuberculosis Disease in North Sumatra Province

Tuberculosis is an infectious disease caused by Mycobacterium tuberculosis and remains a serious health problem in Indonesia, particularly in North Sumatra Province. The increase in the number of cases from 2021 to 2023 indicates the need for an analytical approach that simultaneously considers spatial and temporal aspects. This study aims to apply the Geographically Weighted Panel Regression (GWPR) method to analyze the development of tuberculosis and identify the factors that significantly contribute to its spread in 33 regencies/cities in North Sumatra. Based on the results of the Chow and Hausman tests, it was found that the Fixed Effect Model (FEM) is the most suitable panel data approach to use before applying the GWPR model. The analysis shows that the three variables that most significantly influence the number of tuberculosis cases spatially are the population size, gender (male), and age ≥14 years. The application of GWPR with adaptive bisquare weighting resulted in the best model with an AIC value of 1141.567 and a coefficient of determination (R²) of 0.99578, indicating that GWPR is a highly effective approach for analyzing the spatial spread of infectious diseases such as tuberculosis. The GWPR model is able to explain the spread of tuberculosis more accurately compared to FEM because GWPR can capture the varying influence of each variable in each region, whereas FEM only produces a single coefficient value that applies to the entire area without considering the existing spatial variations.