Date of Publication : 29th June 2024

Abstract:PCOS, or polycystic ovarian syndrome, is categorized as a serious health issue that affects women worldwide. Early PCOS diagnosis and treatment lowers the risk of long period consequences, such as elevated risk of diabetes that of type 2 and gestational diabetes. Consequently, the healthcare systems will benefit from a reduction in the issues and consequences associated with PCOS via efficient and early detection. Recently, promising achievements in medical diagnostics have been demonstrated via Machine Learning (ML) and ensemble learning. Our research's primary objective is to offer local and global model explanations that will guarantee the established model's efficacy, efficiency, and reliability. Various machine learning models may be used as feature selection techniques to obtain the best model and optimum feature selection. To increase performance, stacking ML models—combining meta-learner with the best base ML models—is suggested. ML models are optimized by Bayesian optimization. An 80:20 and 70:30 ratio-splitting benchmark PCOS dataset was used to get the experimental results. In comparison to other models, the outcome shown that utilizing ML with REF feature selection achieved the greatest 100 percent accuracy.

Reference :