PCOS Accuracy Detection Model

Six to twelve percent of women who are of reproductive age have PCOS, which can cause metabolic problems and infertility. Diagnosing PCOS is difficult because of its wide range of symptoms and lack of accepted diagnostic criteria. In order to increase diagnostic accuracy, this work uses exhaustive grid search to optimize the Random Forest and XGBoost classifiers. Both models were assessed on important metrics following hyperparameter adjustment, with XGBoost outperforming Random Forest. The findings demonstrate the promise of machine learning for early and precise PCOS diagnosis, resulting in better patient outcomes and more efficient healthcare delivery. There are suggestions made for both therapeutic application and additional study The manifestations are very diverse, and there is no strict test or diagnosis for this, therefore making it really tough to diagnose PCOS. Recognition and prediction of PCOS have recently presented itself as the most promising machine learning tool. This paper proposes a new diagnostic system and predictive tool for PCOS, based on a set of clinical and metabolic parameters, and a novel approach toward machine learning algorithms. Three different algorithms are utilized in the system: SVM, RF, and LR. In this work, a three-algorithm-based system is proposed, which is tested using the holdout validation method, where its accuracy rate was found to be 89.02%. The evidence presented herein thus indicated potential early detection and prediction through the application of machine learning for PCOS.