Investigation of Polycystic Ovary Syndrome Diagnosis Using Machine Learning Approaches

This paper investigates machine-learning approaches for predicting polycystic ovary syndrome (PCOS) using feature engineering and feature selection. It considers CatBoost, XGBoost, LightGBM, AdaBoost, and Random Forest models, together with several feature-selection strategies. According to the abstract, AdaBoost trained on ten features selected through Random Forest feature importance and Highest Correlation achieved the best test accuracy. However, the supplied abstract does not report the dataset size, the actual accuracy, class balance, sensitivity, specificity, calibration, or external validation. The work therefore appears more relevant as a comparative modeling study than as evidence of clinical readiness.