<![CDATA[Ovary Syndrome (PCOS) is an endocrine-related condition predominantly affecting women during their childbearing years who experience delayed diagnosis due to the limitations of conventional methods that require laboratory tests and imaging procedures that are relatively costly and time-consuming. This study develops a PCOS classification model based on a clinical dataset of 541 patients with 42 clinical attributes using the random forest algorithm with Recursive Feature Elimination (RFE) feature selection and an Explainable AI (XAI) approach. The research pipeline comprised several sequential stages: problem identification, data collection, preprocessing, data splitting, feature selection, model training and testing, evaluation, and SHAP-based explainability analysis. Performance was evaluated using Accuracy, Precision, Recall, and F1-score, and compared between two models, namely RF+CF and RF+RFE, where RF+RFE was identified as the best-performing model. The XAI approach using SHAP (SHapley Additive exPlanations) was applied to identify and explain the contribution of clinical variables to the classification results. The best model, RF+RFE, achieved an accuracy of 92.66%, precision of 93.75%, recall of 83.33%, and F1-score of 88.24%, demonstrating superior performance compared to RF+CF. As this study relies on a single dataset, broader validation across multiple centers is recommended before clinical deployment. This model is intended as a screening-support approach and has not been validated as a clinical diagnostic tool. The findings are anticipated to serve as a foundation for building data-driven early screening tools and clinical decision-making support systems.]]>
Copyrights © 2026
