Modelos de regressão para incidência de HIV/AIDS na população dos estados da região Norte do Brasil

Abstract

AIDS is a notifiable disease that has no cure, but modern medicine can extend the life of the patient with drugs that minimize the effect of the disease on time, giving better quality of life for the person carrying the virus. In Brazil, cases of AIDS have stabilized at around 40,000 per year. The treatment in the National Health System and prevention campaigns enabled the country to stabilize the disease, but in the poorest states in public health structure, growth rates remain high, as is the case in northern Brazil. In this sense, this study aimed at using regression models for modeling of AIDS cases in the population of the States of Northern Brazil. The methodology used consisted of using linear and nonlinear models with use of CurveExpert and Statistica programs, the models were subjected to residual analysis that found normality, independence and homoscedasticity, and outlier analysis. The evaluation of the predictive capability was made via BIC and AIC criteria. Data were obtained from the DATASUS portal, we collected all AIDS data from 1984 to 2014. The main results stands out that for the northern region, the model that best fit the data was the non-linear model logistics designing future stability of AIDS cases for the North estimated at 4700 new cases annually.