Development of a machine learning model to classify polycystic ovarian syndrome.

BackgroundOne of the main causes of infertility among women nowadays is Polycystic Ovarian Syndrome, or PCOS.

Objectives

A decision support strategy for supporting medical specialists through PCOS monitoring is presented in the suggested work. A feature selection model that is based on wrapper categorization is used in this work. The performance of the classifier may be impacted by the existence of redundant features.Method: In order to address these issues, the PCOS is classified using the Machine Learning (ML) Extreme Gradient Boosting (XGBoost) classifier, while the best features are found using the Adaptive Tunicate Search Algorithm (ATSA). By using the number of optimized features as the fitness function, the suggested ATSA increases the capabilities for exploration and exploitation.ResultsUsing the PCOS dataset from the Kaggle repository, the experimental demonstration produced better accuracy and precision than the traditional feature selection methods, achieving 97.5% and 95.7%, respectively.ConclusionThe suggested ATSA + XGB model's scalability was demonstrated by all experimental analyses.