(2024-03-07) Santos, Daniel Ramos Correia dos; Albuquerque Júnior, Gabriel Alves de; http://lattes.cnpq.br/1399502815770584
Cardiovascular diseases and diabetes represent significant challenges for public health, requiring effective diagnostic and prevention approaches. This work proposes an approach based on machine learning models to support these processes. Using a database from the IBGE national health survey, the study investigated how different variables affect the detection of these diseases. Using algorithms such as Random Forest, XGBoost and SVM, predictive models were developed. The results demonstrated an accuracy of 71.96% for the Random Forest algorithm in classifying patients with cardiovascular diseases and 72.26% in classifying patients with diabetes. Analysis of the most influential variables was also carried out using the SHAP method, which revealed some insights into the data.
