Diagnosis of Polycystic Ovary Syndrome using Machine Learning Algorithms
Vineela Madireddy, Shaik Rehmat Sultana, Muddasani Ashvitha, Yadasu Ajay · 2025
This Mission focus on approaches for the diagnosis of polycystic ovarian syndrome in women. By using machine learning methods from a publicly available datasets like Kaggle, this study main target is to improve the accuracy and efficiency of the diagnosis. PCOS has become a common health problem among women of reproductive age, significantly affecting fertility, overall health of women. Older diagnostic methods rely on clinical treatments and ultrasound, which takes more time and prone to errors. To solve these challenges and issues this research proposes an automated approach leveraging ML algorithms for early and accurate PCOS detection. The dataset comprises of 43 attributes from 541 women, including 177 diagnosed with PCOS. Feature selection technique such as univariate analysis, help identify the most influential predictors, with the ratio of follicle- stimulating hormone (FSH) to luteinizing hormone (LH) emerging as a key factor. Machine learning models, including Random Forest, Logistic Regression, Gradient Boosting, and a hybrid Random Forest-Logistic Regression (RFLR), are trained on selected features. RFLR achieves the highest testing accuracy of 91.01% with a recall of 90% using 40-fold cross-validation.