Detection and Classification of Polycystic Ovary Syndrome using Machine Learning-Based Approaches

polycystic ovarian syndrome, or PCOS, is categorized as a severe health problem that impacts women globally. Early detection of PCOS reduces the chances of long-term effects, including a higher risk of gestational and type 2 diabetes. Consequently, healthcare systems will benefit from a reduction in the issues and complications associated with PCOS through efficient and early detection. For early diagnosis and prediction of PCOS, we have applied machine learning techniques on a dataset comprising 541 patients from India. Feature selection techniques along with machine learning models such as support vector machines, random forests, decision trees, naïve bayes, logistic regression, Gradient boosting, Catboost and Adaboost models were used for PCOS prediction. After comparing the results of each classifier, it was found that the Catboost Classifier had the highest and most reliable precision, accuracy, recall and F1-score, which were 95%, 93%, 96%, and 96%.