Identification of Polycystic Ovarian Syndrome through Clinical Metrics and Ultrasonographic Imaging Analysis

Polycystic ovarian syndrome (PCOS) is an endocrine disorder which is characterized by hormonal dysregulation and it commonly affects women during their reproductive age. The incident of an irregular or entirely missing monthly menstrual cycle can be attributed to a disturbance or imbalance in the complex interaction of hormonal secretions and their control mechanisms within the female reproductive system. The cause of Polycystic Ovary Syndrome (PCOS) remains hard-to-find, lacking a universally acknowledged or defined factor responsible despite ongoing research efforts aimed at clarifying its origin and pathophysiology. PCOS symptoms like weight gain, facial hair, acne, and irregular cycles impede fertility. Weight management and preliminary intervention may reduce diabetes and heart risks, yet present techniques come up lacking in Early identification and prediction. We suggest a method using machine learning algorithms to predict Treatment for Polycystic Ovary Syndrome effectiveness based on Key criteria, striving for early detection using a Kaggle dataset. The paper incorporates the utilization of two different datasets aimed at facilitating the process of classification. Phase one entails identifying the primary features that contribute to PCOS and forecasting the onset of the disorder, where among different algorithms Random Forest gave the best accuracy. Phase two encompasses the utilization of an image dataset wherein the classification of images into infected or not infected of PCOS is carried out through the implementation of Convolutional Neural Networks (CNNs).