Title: Advancing Breast Cancer Prediction with State-of-the-Art Boosting Ensemble Learning: A Systematic Review of Performance, Interpretability, and Clinical Impact
Authors: Charity Segun ODEYEMI, Muibat Temitope BOLANLE and Olumhense Benedict ADOGHE
Volume: 10
Issue: 6
Pages: 29-41
Publication Date: 2026/06/28
Abstract:
Breast cancer has been adjudged the leading cause of cancer-related mortality worldwide, necessitating accurate and early diagnostic approaches to improve clinical outcomes. Recent advances in machine learning have highlighted the effectiveness of boosting ensemble techniques-such as XGBoost, LightGBM, CatBoost, and AdaBoost-in enhancing predictive performance for breast cancer diagnosis. This study presents a systematic review of boosting-based models for breast cancer prediction, conducted using the PRISMA 2020 framework to ensure a rigorous and transparent selection process. A total of 46 peer-reviewed studies published between 2018 and 2026 were analyzed, covering diverse datasets including clinical, imaging, and multi-omics sources. The findings indicate that boosting algorithms consistently achieve high classification accuracy (above 97%) and AUC values exceeding 0.98, outperforming traditional machine learning models. Hybrid frameworks integrating deep learning and boosting further improve performance, particularly in high-dimensional imaging data. Additionally, the integration of Explainable Artificial Intelligence (XAI) enhances model interpretability and clinical trust. However, challenges such as data imbalance, limited generalizability, and high computational complexity persist. Future research directions emphasize multimodal data fusion, federated learning, and edge-based deployment. This review provides a comprehensive synthesis of current trends, challenges, and opportunities, offering valuable insights for advancing AI-driven breast cancer prediction systems.