An Analysis of Polycystic Ovarian Syndrome (PCOS) Among College Students in Ahilyanagar, Maharashtra, Based on Different Statistical Models

Abstract

Polycystic Ovarian Syndrome (PCOS) creates a hormonal imbalance among females, compromising their reproductive health. Apart from disruption in reproductive health, this syndrome causes disturbances in various attributes of an individual, such as psychological, metabolic, and circadian rhythm. It also increases the risk of other diseases such as non-alcoholic fatty liver and cardiovascular disorders. The existence of PCOS tendency for a long time causes Infertility; hence, its early detection will help in managing its aetiology. Detecting PCOS is difficult due to the complexity of PCOS pathogenesis, so the Rotterdam criteria are the gold standard. The present study aims to understand the prevalence of PCOS among college-going girls in Ahilyanagar city (formerly Ahmednagar). The study included 1350 females between the ages of 16 to 35. 111 females were clinically diagnosed with PCOS, and a prevalence of 8.22% in Ahilyanagar city, Maharashtra, was detected. Based on the chi-square analysis, PCOS females were associated with hyperandrogenism, oligomenorrhea, and skin darkening; Acanthosis nigricans, an indicator of insulin resistance, was also positive. A study of lifestyle parameters showed a positive association with PCOS. Four statistical models are proposed for the diagnosis of PCOS: Random Forest, Naïve Bayes, Support Vector Machine, and Logistic regression model. Naïve Bayes is the most effective model for the collected data, with an accuracy of 91.48 and sensitivity of 0.9090, followed by Logistic regression with an accuracy of 91.11 and sensitivity of 0.9005.