Multivariate Bayesian Spatio-Temporal Modeling of Malaria Incidence and Mortality in Kenya

In Kenya, malaria remains a major public health challenge, affecting a substantial proportion of the population and exhibiting pronounced spatial and temporal variability in transmission patterns. A comprehensive understanding of the spatial and temporal distribution of malaria incidence and mortality is therefore essential for the design and implementation of effective, targeted intervention strategies aimed at reducing disease burden and preventing malaria-related deaths. This study aimed to develop and apply a Multivariate Bayesian Spatio-Temporal modeling framework incorporating skew distributions to jointly analyze the spatio-temporal distribution of malaria incidence and mortality in Kenya. This proposed approach allowed for the explicit characterization of shared spatial structures and temporal trends, as well as the dependence between incidence and mortality across different counties and time periods. Parameter estimation was conducted using the Markov Chain Monte Carlo (MCMC) algorithm, which enabled sampling from the posterior distributions and facilitated robust statistical inference under uncertainty. The performance of the proposed model was assessed using established Bayesian model evaluation criteria which included the Widely Applicable Information Criterion (WAIC), the log pointwise predictive density (lppd), and the effective number of parameters (pWAIC). These metrics were used to evaluate model fit, predictive accuracy, and complexity, ensuring a balanced assessment of model performance. The results indicated that the multivariate Bayesian spatio-temporal model effectively captured the underlying spatial and temporal dependencies in malaria incidence and mortality across Kenya. The model successfully identified variations in risk across the Kenyan Counties and time periods, demonstrating its capacity to represent intricate malaria dynamics. Thus, this study findings highlight the utility of multivariate Bayesian spatio-temporal modeling as a powerful tool for understanding malaria transmission patterns and for informing evidence-based, spatially targeted malaria control and prevention strategies in Kenya.