MONITORAMENTO ONLINE DA DENGUE: USANDO O GOOGLE PARA PREDIZER EPIDEMIAS

Main Article Content

Rafael IZBICKI
Letícia Octaviano da CRUZ

Abstract

Unfortunately, oficial reports on the development of dengue in Brazil take a long time to be released. This creates an environment where policy makers do not have enough accurate information they can use to improve how they act in order to prevent dengue to spread. In this paper we show how data collected in  real time from Google Trends can be used to predict the current number of dengue cases in the state of 
Sao Paulo. In order to estimate the number of new cases as a function of time, we use the least square method, lasso and random forests, having the search volume of several keywords as covariates such as "tratamento dengue", "sintomas dengue", and  "febre dengue" in the models. The least square method presented better predictions and gave reasonable estimates for up to eight months after the last release of an ocial report. This allows authorities to take necessary actions to prevent dengue from spreading in a much cheaper and efective way.

Article Details

How to Cite
IZBICKI, R., & CRUZ, L. O. da. (2018). MONITORAMENTO ONLINE DA DENGUE: USANDO O GOOGLE PARA PREDIZER EPIDEMIAS. Brazilian Journal of Biometrics, 36(3), 512–526. https://doi.org/10.28951/rbb.v36i3.227
Section
Articles