Study of the Temporal Propagation of Arboviruses in the Region of Recife-PE: Analysis of Climatic Influence using the SIR Model and Recurrent Neural Networks
DOI:
https://doi.org/10.55592/cilamce.v6i06.10164Palavras-chave:
Reinforcement learning, Epidemiological Models, Climate changeResumo
The occurrence of disease outbreaks, especially Dengue, Zika and Chikungunya, is on the increase throughout Brazil, and is currently a significant concern for the Recife-PE region due to the high temperatures. This directly affects public health and the population's quality of life, as well as the city's economy and education. To deal with this challenge, public policies are being implemented, such as awareness campaigns, inspections, and case reports. However, underreporting is a common problem due to the lack of demand for health services and difficulties in accessing medical care, which is reflected in official municipal data. In addition, there are gaps in the treatment of the specificity and cause of the problem and its mitigation. To overcome this inaccuracy, dynamic models such as the SIR model have been widely used in epidemiology. This model, which is based on differential equations, describes the temporal evolution of the susceptible, infected, and recovered classes. In Recife, the ambient temperature shows a strong positive correlation with the infection rates of cases of the diseases, which leads to the intensification of prevention campaigns during the summer. A study carried out in 2022 used data from the National Institute of Meteorology (INMET) to adjust a trigonometric function and analyse the seasonal influence of climate on infection rates, applying the SIR model. In addition, monthly iterations were carried out using the Runge Kutta numerical method in Simulink software. To improve the prediction and qualification of endemic disease cases, long-term memory modelling was used with a Recurrent Neural Network (RNN), validated based on available epidemiological data, obtaining an RMSE error metric of around 0.8 for the three diseases assessed.