•  
  •  
 

Abstract

Polycystic ovarian syndrome (PCOS) is a common endocrine disorder affecting women of reproductive age worldwide, characterized by excess production of androgens. This can result in ovarian abnormalities and a range of associated health risks, including infertility, heart issues, diabetes, and uterine cancer. However, the diagnosis of PCOS can be challenging due to the varied symptoms in different women and the time and cost involved in biochemical tests and ovarian scanning. To address this, researchers have proposed a method that predicts the likelihood of PCOS based on a minimal set of criteria, including weight, BMI, cycle length, and hormone levels. Using five machine learning algorithms, they tested the method on a dataset of 541 patients and found that the Random Forest and Support Vector Machine models had the highest accuracy in predicting PCOS. Such a system could aid in early detection and encourage individuals at risk to seek medical attention. Dataset is split into a 70/30 ratio, meaning that 70% of the dataset’s data are used to train the model and 30% are used to test it. In this paper, we suggested a novel stack model with a 90% accuracy that is composed of four machine learning classifiers: Random Forest, Support Vector Machine, Naive Bayes, and Logistic Regression. Testing data accuracies for the models of Logistic Regression, Random Forest, Support Vector Machine, K Nearest Neighbor, Naive Bayes, Stack Model are 88%, 91%, 90%, 69%, 86% and 90% respectively. As a result, the models with the highest accuracy on the testing data are the Random Forest model and Stack Model.

Download

Share

COinS