Modelling of Dengue Hemorrhagic Fever Disease in Semarang City Using Generalized Poisson Regression Model

Abstract

Dengue Hemorrhagic Fever (DHF) is an infectious disease that can be life- threatening within a relatively short period of time and can be fatal if not promptly treated. DHF in Indonesia ranks second as a dangerous seasonal disease. DHF remains a serious issue in the Central Java Province, particularly in Semarang City. The cases of DHF can be modeled using a Poisson regression model due to the characteristics of DHF cases, which involve count data with small occurrence probabilities. The Poisson regression model assumes equality between the mean and variance (equidispersion). However, the application of the Poisson regression model often encounters violations of the assumption of excessive variance (overdispersion), which necessitates addressing the violation, and one possible approach is to use the Generalized Poisson Regression model. Based on the analysis results, the Generalized Poisson Regression model could handle the overdispersion because the ratio of Pearson Chi-Square by degrees of freedom was 0.976, approaching a value of 1. It has also been proven to be more suitable for evaluating factors influencing the number of DHF cases, as it has a lower AIC value compared to Poisson models, with a value of 123.64. The variables that were found to have an impact on DHF cases in Semarang City based on the Generalized Poisson Regression model are the number of larval habitats (X1), the number of hospitals (X2), population density (X3), and the number of healthcare workers (X4).