Tehran University of Medical Sciences Frontiers in Biomedical Technologies 2345-5837 0 0 2026 07 07 1462 1462 Mojtaba HajiHasani Assistant Prof at Amol University of Special Modern Technology 2025 09 07 2025 12 30 Objective: Metabolic syndrome (MetS) significantly increases the risk of cardiovascular disease and is characterized by a combination of atherogenic dyslipidemia, central obesity, and hypertension. This study investigates additional factors contributing to MetS and their relevance by employing advanced analytical techniques. Methods: Specifically, we utilized ensemble learning, a supervised learning approach that combines multiple individual models to enhance predictive performance. Input factors were ranked through feature selection techniques, while unsupervised methods were employed to mitigate biases associated with traditional labeling based on textbook criteria. An autoencoder was trained to reduce dimensionality without requiring prior knowledge of the data. Results: Our findings reveal an impressive accuracy of 98.22% achieved by the gradient boosting classifier when evaluating 23 factors. Contrary to the expectation of identifying only five main factors as per the ATP III guidelines, our analysis highlights HOMA-IR, BMI, age, and cholesterol as critical predictors of MetS. Furthermore, the autoencoder's compressed representation uncovers a distinctive three-state classification, rather than the conventional binary classification of MetS presence or absence. Two clusters showed a strong correlation with MetS and non-MetS classifications, while the third cluster represents instances on the boundary of MetS diagnosis. Conclusion: It is shown that obesity followed by high blood sugar and hypertriglyceridemia, insulin resistance (HOMA-IR), hypertension, to a lesser extent, age and cholesterol appeared to be the most essential features of the MetS prediction. Additionally, discovery of a distinct third cluster introduces a perspective on MetS, suggesting the potential for developing an early-staged nuanced patient management strategy. https://fbt.tums.ac.ir/index.php/fbt/article/view/1462 https://fbt.tums.ac.ir/index.php/fbt/article/download/1462/573