PCOS Disease Classification Using XGBoost Algorithm and Genetic Algorithm for Feature Selection

Abstract

Polycystic Ovary Syndrome (PCOS) is an endocrine disorder that often occurs in women of reproductive age, with a global prevalence of 10-16%. The diagnosis of PCOS is still a challenge due to the uncertainty of the cause, which can worsen the patient's condition due to delayed detection. This study aims to develop a classification model to detect PCOS using a combination of SMOTE algorithm, genetic algorithm, and XGBoost. The dataset used is a public dataset from Kaggle entitled "Diet, Exercise, and PCOS Insights". A genetic algorithm was used to select the best 15 features, while SMOTE was applied to handle data imbalances. XGBoost is used for classification with a model accuracy of 82.86% and an F1-score of 88% for the PCOS negative class and 70% for the PCOS positive class. The results show that combining these algorithms can improve the accuracy of predictions and offer more efficient diagnosis solutions. This research is expected to contribute to developing early diagnosis methods for PCOS.