Malaria incidence in a region fluctuates over the year due to seasonal factors such as rainfall and temperature. In this project, we use health facility case data to determine the temporal trends across Madagascar. In the absence of the complete set of locations and catchment population estimates, we focus on location-specific monthly case proportions.
A spatio-temporal log-linear regression model is used to relate the empirical monthly case proportions to environmental covariates as well as to account for unexplained, residual spatio-temporal effects. To visualise the results and the associated uncertainty, we apply a seasonal feature extraction algorithm to the posterior samples from the fitted model.
The derived maps of the transmission peaks, lengths as well as start and end of the transmission seasons are useful for policymakers when they plan intervention strategies such as indoor residual spraying. This work is collaboration with many colleagues at the Malaria Atlas Project, Imperial College London, Institu Pasteur de Madagascar and the National Malaria Control Programme in Madagascar. The related paper has been published at BMC Medicine. Sample code and data are available at an online repository.