Keywords: Malaria, Spatiotemporal, Recommendation engine
The malaria research community has made great strides in mapping disease prevalence, modeling its transmission, developing and testing effective interventions such as bed nets. We propose to build on this work to develop a real-time recommendation engine (RE) for precision interventions to help policy-makers decide how to allocate their limited resources. Because intervention effects can vary spatiotemporally and spillover into neighboring regions, targeted interventions that are optimized to maximize efficacy over the entire spatial domain are key to improving eradication efforts. The truly optimal strategy for spatiotemporal resource allocation is a massively complex mathematical function of many geographic and epidemiologic inputs, requiring extensive computation, and is essentially a black box that generates no new knowledge. In contrast, we propose an approximate strategy that lays bare the contribution to the RE of various risk factors, permits straightforward computation, and is amenable to visualization, scrutiny, and stakeholder input. These features are essential for the RE to be adopted by malaria researchers and policy makers.