Artificial neural networks applied in the correlation between dengue deaths, self-medication and abiotic factors in Goiânia-Goiás

Abstract

This article aims to verify in an unprecedented manner by means of an explanatory study the possible correlations that still not well understood between dengue deaths and abiotic factors, like climatic conditions and the influence of self-medication, it was analyzed for the city of Goiania-Go Brazil. Thus, it was collected government data for these variables between 2005 and 2015, and then were performed a multidimensional approximation on these data by the Artificial Neural Networks (ANN), using the Multilayer Perceptron architecture, optimized by the algorithm of Levenberg-Marquardt, and it was performed the design and studied of six topologies. For calculate the significance of the correlation between the input variables of the ANN was proposed modifications to Profile method. The results showed that there is a large influence of self-medication and precipitation as the main striking in deaths, the temperature were classified as moderate significance. Thus, it was concluded that the ANN could be applied satisfactorily in modeling epidemiological problems.