查看文章

Dengue is a mosquito-borne viral disease that infects over 400 million people per year worldwide, resulting in vast disease burden. Dengue is transmitted primarily by mosquito species Aedes aegypti and Aedes albopictus, and is endemic to tropical regions; however, wherever these mosquitoes expand their habitats, the disease can spread. Forecasting dengue incidence is inherently challenging because of the difficulty of tracking and estimating mosquito populations. We adopt a data fusion approach that incorporates environmental variables as proxies for mosquito density at the municipality level of resolution. We apply penalized regression models to historic weekly dengue data for representative tropical and temperate metro areas in Brazil. We compare results from varied rolling retrocast windows, and quantify the error using Mean Squared Error and R2. We improve our retrocasts by incorporating lag times to …