Kernel biresponses multipredictors nonparametric regression for modeling maternal and infant mortality rates in East Java, Indonesia

The rates of infant and maternal mortality (IMR and MMR) serve as important public health indicators worldwide. In Indonesia, East Java Province significantly contributes to these statistics, underscoring the need for targeted interventions. This study explores the key determinants of MMR and IMR, including the density of health facilities, household lifestyles, early marriage rates, poverty levels, and antenatal care coverage. The correlation between MMR and IMR suggests interconnected health outcomes requiring deeper analysis. A kernel estimator in biresponse multipredictor nonparametric regression (KBM-NR) is utilized to address these complexities, with the Weighted Least Squares method applied for estimation. Among seven kernel functions tested, the Triweight Kernel function is identified as the most optimal, achieving a Generalized Cross-Validation value of 71.2352 and an R-square value of 97.116%. These findings demonstrate the model's strong predictive capability and potential for informing strategies to reduce maternal and infant mortality rates in East Java.