Phenotype-Specific Lifestyle Prediction for PCOS Using Machine Learning Multi-Class Classification and SHAP Explainability

This study addresses the heterogeneity in Polycystic Ovary Syndrome (PCOS) phenotypes by developing a machine learning framework to predict four specific phenotypes using non-invasive lifestyle and symptom data from 267 patients. Utilizing five classifiers—Support Vector Machines, Extreme Gradient Boosting, Random Forest, Logistic Regression, and K-Nearest Neighbours—the models achieved high accuracy (≥ 98%), with XGBoost and Random Forest achieving perfect separation. SHAP analysis identified cycle_length as the most significant predictor across all models, underscoring its clinical relevance as a biomarker for PCOS. The study demonstrates that explainable machine learning can facilitate phenotype-specific lifestyle recommendations, potentially improving PCOS management.