摘要：Malaria is still a fatal disease in many countries around the world. Establishing measures to control and prevent this disease has been a worldwide concern since 1950, when the World Health Organization launched a Malaria Eradication Plan. In Brazil, malaria was eliminated in much of the territory, but resisting in an area known as the Legal Amazon. That said, the main scope of this work is to develop statistical control charts that consider the temporal dependency structure in the data and are suitable for the current and future monitoring of malaria cases, in order to detect possible outbreaks or epidemics in states in the Legal Amazon region. The tools presented here could also collaborate in directing control actions and combating the spread of the disease. In particular, we intend to: (i) build a statistical model to predict the occurrence of cases of the disease, which considers the existence of a possible temporal dependency structure between the collected data; (ii) use the Statistical Process Control (SPC) techniques, notably the control charts, to monitor (separately for each form of the disease) cases of malaria of the types Plasmodium Vivax, Plasmodium Falciparum and Plasmodium Mista in the Amazon region; (iii) establish epidemic thresholds based on the obtained control charts. Considering data from 2013 to 2017, this study revealed, among others, that the generalized autoregressive and moving average models with Negative Binomial distribution (Negative Binomial GARMA models) were more efficient, fitting better, compared to Poisson GARMA models, due to the overdispersion existing in the analyzed data.
