An Integrated Bayesian Modelling Approach for Predicting Mosquito Larval Habitats

Malaria is the leading cause of death in Kenya highlands. Malaria is a vector-borne disease and mosquito is the vector that transmits the parasites from infected people to others. Controlling mosquito larval habitats is important in eradicating malaria. Modeling the spatially distributed mosquito larval habitats is challenging. This study explores a mosquito habitat modeling approach, which integrates a Bayesian modeling method with Ecological Niche Factor Analysis (ENFA). ENFA transforms original environmental variables into niche factors. These factors are extracted to maximize the variations between the global space, which represents the general conditions of an area, and the focal space, which represents habitats. The integrated method is compared with the Bayesian method without ENFA. Both methods accurately predicted the habitats approximately 70-80%. Although the two methods have similar prediction accuracies, the integrated approach produces more ecologically justified spatial distribution of larval habitats. The map derived from this estimation has important implications for malaria control and eradication.