Dengue Vector Population Forecasting Using Multisource Earth Observation Products and Recurrent Neural Networks
Oladimeji Mudele, Alejandro C. Frery, Lucas Zanandrez, Álvaro Eduardo Eiras, Paolo Gamba
Abstract
This article introduces a technique for using recurrent neural networks to forecast <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Ae. aegypti</i> mosquito (Dengue transmission vector) counts at neighborhood-level, using Earth Observation data inputs as proxies to environmental variables. The model is validated using <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in situ</i> data in two Brazilian cities, and compared with state-of-the-art multioutput random forest and k-nearest neighbor models. The approach exploits a clustering step performed before the model definition, which simplifies the task by aggregating mosquito count sequences with similar temporal patterns.