Modelagem e predição espaço-temporal dos casos de dengue utilizando processo pontual de Cox log-Gaussiano

Abstract

Dengue fever is an infectious viral disease that has caused great concerns for public health in Brazil in recent years. Among the Brazilian states that suffer the most from this disease, the Minas Gerais’ state stands out. Knowing the behavior of the dengue virus in relation to its way of propagation in places and times of great incidence is of paramount importance in order to reduce the number of these occurrences. The statistic can be an important tool for combating the disease, mainly using methods and techniques that consider time and spatial location relevant informations in the analysis. In this work, it describes the behavior of dengue using the Cox log-Gaussian model for space-time in order to identify regions where the risk of disease is high. Analysis considered notifications of the cases occurred in the Três Corações city during the years 2010 to 2015. It was verified through descriptive analysis, modeling and prediction that the period with large number of occurrences happens between February and June. Areas that were at high risk of the disease included the following neighborhoods: Peró Um, Peró Dois, Santana, Parque São José, Vila Lima, Loteamento Bela Vista, Odilon Rezende, Vila Gesse, Monte Alegre, Centro, Jardim Santa Tereza, Cotia, Vila Fernão Dias, Vila Santo Afonso, Jardim Califórnia, Jardim Paraíso, São Jerônimo and Cinturão Verde. With these results, we observed a certain proximity among days and these neighborhoods too, which reflects the presence of clusters of dengue’s cases both in time and space, typical feature of this disease.