Comparative Analysis of Machine Learning Algorithms in Diagnosis of Polycystic Ovarian Syndrome
Malik Mubasher Hassan, Tabasum Mirza
Abstract
Artificial intelligence can be used in healthcare systems for diagnostic purposes to handle large amounts of clinical data with much accuracy and precision. One of the commonest health issue found in the young women is Polycystic Ovarian Syndrome (PCOS) and it is basically a complex health disorder affecting women of reproductive age group that can be diagnosed based on clinical symptoms like increased body mass index, elevated hormone levels, hair loss, acne, skin darkening, hirustism, cycle length, endometrial thickness, high blood pressure levels, etc. Correct diagnosis is the baseline of any proper treatment and in this research paper we are using machine learning approaches like Support Vector Machine, CART, Naive Bayes Classification, Random Forest and Logistic Regression to diagnose PCOS based on the clinical data of patients. The results were analyzed and performance of the algorithms was validated on the basis of accuracy, precision, recall, F-statistics, and Kappa Coefficient. The validation metrics indicate the highest i.e. 96% accuracy of the Random Forest algorithm in the diagnosis of PCOS on giving data.