Bayesian Spatio-Temporal Model Simulation of Climatic Influences on Malaria-Typhoid Dynamics

Abstract

Typhoid and Malaria are among the world’s biggest health challenges, with the Sub-Saharan Africa being the most affected. Although typhoid and malaria have different causes and modes of spread, they both have similar symptoms. Despite historical attempts to model the co-occurrence of malaria and typhoid have achieved some success, they often overlooked the inter-connectedness between climatic factors, spatio-temporal disease transmission dynamics, and co-infection. The aim of this study was to develop a Bayesian hierarchical spatio-temporal model for the joint analysis of malaria and typhoid, which includes both specific and shared spatio-temporal effects under an influence of climatic factors. Parameters for variables such as temperature, precipitation and humidity were estimated under a simulation study using Bayesian inference through the R-INLA approach which serves as an alternative to MCMC since it tackles even intractable integrals. After developing the model, we assessed the properties of the estimators. Plots alongside statistics such as RMSE, variance, Shapiro, mean Z, Skewness, Kurtosis among others were done. The estimator was unbiased, asymptotically normal, asymptotically efficiency and consistent. For increasing observations beyond 1,000 the bias, standard deviation and root mean squared errors (RMSE) decrease hence the sample size was sufficient. Thanks to the simulations, it was observed that temperature had a high impact on the co-infection than other variables while lower precipitation was associated with increased risk. The structured spatial effect revealed a bigger influence on the disease co-infection than other spatial and temporal components.