Lifestyle-Based Prediction of Polycystic Ovary Syndrome (PCOS) Using Machine Learning

Polycystic Ovary Syndrome (PCOS) is a pervasive endocrine disorder affecting reproductive-age women globally. Early detection is critical to mitigating long-term complications such as infertility, type 2 diabetes, and cardiovascular diseases. However, traditional diagnostic protocols often necessitate invasive clinical procedures and extensive laboratory diagnostics, which are not always accessible or affordable. This paper presents a non-invasive, predictive framework for PCOS detection centered on modifiable lifestyle and anthropometric indicators. By leveraging the Random Forest machine learning algorithm, we analyze a dataset comprising critical lifestyle variables, including Body Mass Index (BMI), physical activity frequency, dietary habits, and sleep quality. Our methodology includes data preprocessing and feature selection to identify the most significant clinical predictors. Experimental results demonstrate that this approach achieves high predictive accuracy, suggesting that a Random Forest-based screening tool can serve as a viable, low-cost primary assessment tool. This study provides a scalable approach for early risk identification, enabling timely lifestyle interventions and personalized healthcare recommendations for women at risk of PCOS.