Performance Evaluation of various ML Algorithms for PCOS Diagnosis

PCOS is a common endocrine disturbance leading to anovulation and subsequent severe health disorders such as cardiovascular events, type 2 diabetes, and infertility. An early and correct diagnosis is crucial to managing the disease and optimizing clinical outcomes. Traditional methods used to diagnose PCOS involve physical examinations and hormone testing but are not always conclusive, particularly at an initial stage. One technology that could substitute exploring large datasets and identifying magnified patterns that might assist in predict disease is machine learning. In this paper, we examine whether numerous ML algorithms can recommend the possibility for women to have PCOS. We use a dataset from a Google Collaboratory in which our features differ from the traditional diagnostic criteria for PCOS. This is the distinctive part of the study that enables us to test the possible advantages and disadvantages of including this extra information in the forecasting models. We will test the efficacy of such characteristics in determining the precise PCOS patients across a variety of classification models. The result of the current research will contribute to the growing body of evidence suggesting that machine learning may be used to identify diseases sooner, and promote women better manage their health.