Bayesian modelling of dengue incidence with climatic drivers: comparing fixed-effects, nonlinear and dynamic approaches

Abstract

Climatic variability plays a critical role in shaping dengue transmission dynamics, yet empirical findings remain inconsistent across studies. Divergent conclusions regarding the associations of temperature, relative humidity, wind speed, air pressure, precipitation, number of rainy days, and sunshine duration with dengue incidence often stem from unmodelled interactions and methodological limitations. To address these challenges, this study applies a Bayesian modelling framework to examine the associations between climatic drivers and dengue incidence in Bandung City, Indonesia, using monthly data from 2016 to 2024. We compared fixed-effects, nonlinear and dynamic modelling approaches to evaluate both the direction and magnitude of these associations while addressing overdispersion and potential multicollinearity among predictors. Our findings highlight temperature and relative humidity as the primary climatic variables associated with temporal variations in dengue incidence, with effects manifesting most strongly at a two-month lag. These results underscore the importance of adopting robust Bayesian modeling frameworks to support early warning systems and inform evidence-based public health interventions for dengue control.