APPLICATION OF MACHINE LEARNING ALGORITHMS TO THE PREDICTION OF POLYCYSTIC OVARY SYNDROME (PCOS)

Abstract

PCOS (Polycystic Ovarian Syndrome) is a  prevalent endocrine system disorder that affects roughly 5% to 10% adolescent women. Infertility,cardiovascular diseases,type 2 diabetes etc are the symptoms for the PCOS. By using biochemical,clinical and ultrasonography methods PCOS can be detected. Early diagnosis and treatment will reduce the chance of PCOS.Process of diagnosis includes various aspects like physical examination through symptoms exhibited for a disease,person’s previous medical history ,and various type of medical test. Despite different tools used, the highest accuracy is shown by Logistic Regression  (94.28%). On the other hand, KNN ,Random Forest and naive bayes shows the accuracy performances (87.71%),(92.57%)and (91.42%). For the selection of attributes (the sign and symptoms) machine learning algorithm is used for PCOS patients data which affect the disease condition most ,various classifiers using Logistics Regression comparing the Random Forest, Naive Bayes, K-Nearest Neighbor, have been applied in our dataset and different accuracy parameters also have been used including Confusion matrix , accuracy  to select the best classifier which classify the diseased and non-diseased patients with high accuracy.