Polycystic Ovary Syndrome (PCOS) is a complex hormonal disorder that impacts women. There are still no proven treatments for PCOS, and the exact cause of the condition is still unknown despite much research. For this reason, women’s health benefits greatly from early detection and treatment. Machine learning (ML) based detection methods for PCOS identification have recently shown tremendous performance. When considering the traditional techniques, they are more expensive and take longer time. In this work, a boosting based stacking approach is proposed for PCOS detection. K-fold cross validation is used in this work to offer validation of the PCOS detection model. In order to create this model several boosting algorithms namely AdaBoost, Gradient Boost, Extreme Gradient Boost (XGBoost), CatBoost and LightGBM are used as base learners and Random Forest is used as meta learner. The proposed model when combined with the embedded feature selection technique achieved 96.44% accuracy, 96.16% precision, 96.15% recall, and 96.16% f1-score. Compared to conventional methods, the suggested model shows better prediction performance by utilizing the advantages of stacking and boosting algorithms. This method also shows how ML can help in early and cost effective PCOS detection, which will lead to better patient outcomes.