Predicting the incidence of dengue in Costa Rica using a decision tree model based on climatic and socioeconomic variables

Abstract

Problem: Since 1993, Costa Rica has faced the re-emergence of the disease caused by the dengue virus, and despite the continuous efforts of local authorities to control the vector Aedes aegypti, dengue disease continues to be a problem for the Costa Rican population. Objective: To propose a decision tree model to predict the incidence of dengue in Costa Rica. Method: Quantitative analysis of the incidence of dengue, climatic and socioeconomic variables, by socioeconomic region of Costa Rica, from 2012 to 2018, to perform a predictive model regression decision trees to estimate the incidence of dengue disease per week; as well as its subsequent evaluation with the registered cases of dengue from week 1 to 46 of 2019. Results: The predictive model (RMSE: 5,348) yielded promising estimates for the evaluation period. Conclusions: The added value that predictive models could provide to the control of vector-borne diseases, such as dengue, is demonstrated.