An Efficient Hybrid Stack Ensemble Model For Accurate Breast Cancer Recurrence Prediction Publisher



Davtalab M ; Khosravi H ; Zia E ; Maftoun M ; Shahabi A ; Khademi M ; Atashi A ; Zare O ; Hassannataj Joloudari J
Source: Lecture Notes in Computer Science Published:2026

Abstract

Breast cancer recurrence prediction continues to pose a significant challenge in oncology, as it has a direct effect on treatment approaches and patient outcomes. Conventional predictive techniques frequently struggle to manage the complexities of high-dimensional, imbalanced cancer datasets, hindering their efficacy. This study investigates the use of hybrid machine learning models to address these challenges and enhance predictive precision. The study utilizes an innovative hybrid stacking model (XRS), which combines Random Forest and XGBoost as foundational learners with Support Vector Machine (SVM) serving as the meta-learner, fine-tuned through Bayesian optimization. This systematic methodology adheres to industry-standard processes for data mining, ensuring thorough data comprehension, preprocessing, and model assessment. Experimental findings illustrate the efficacy of the proposed XRS model, achieving impressive performance metrics: accuracy (91%), precision (92%), recall (88%), F1 score (91%), and AUC (96%). These results underscore the potential of hybrid models to deliver more precise, robust, and generalizable predictions, tackling essential challenges in cancer data analysis. By empowering healthcare providers to evaluate recurrence risks with enhanced reliability, this study plays a role in promoting personalized and effective cancer treatment. The encouraging results imply that hybrid models such as XRS could be adopted in real-world medical environments, paving the way for improved clinical decision-making in oncology. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.