Predicting the Occurrence of Metabolic Syndrome Using Machine Learning Models

Abstract

The term metabolic syndrome describes the clinical coexistence of pathological disorders that can lead to the development of cardiovascular disease and diabetes in the long term, which is why it is now considered an initial stage of the above clinical entities. Metabolic syndrome (MetSyn) is closely associated with increased body weight, obesity, and a sedentary lifestyle. The necessity of prevention and early diagnosis is imperative. In this research article, we experiment with various supervised machine learning (ML) models to predict the risk of developing MetSyn. In addition, the predictive ability and accuracy of the models using the synthetic minority oversampling technique (SMOTE) are illustrated. The evaluation of the ML models highlights the superiority of the stacking ensemble algorithm compared to other algorithms, achieving an accuracy of 89.35%; precision, recall, and F1 score values of 0.898; and an area under the curve (AUC) value of 0.965 using the SMOTE with 10-fold cross-validation.