Machine Learning and Deep Learning models for Breast cancer recurrence prediction – A systematic review
Abstract
Early detection of breast cancer recurrence risk is critical for informing treatment decisions and providing personalized patient care. Recent advancements in machine learning (ML) and deep learning (DL) have enabled the development of predictive models that can leverage diverse clinical, imaging, and molecular datasets. For the past few years, numerous machine learning (ML) and deep learning (DL) based studies have been proposed to improve the accuracy of breast cancer recurrence prediction. However, the existing literature varies widely in methodology, data modality, and validation strategies. This study aims to systematically examine current research focused on ML and DL models for breast cancer recurrence prediction, with particular emphasis on methodological trends, data modalities, and model algorithms. Following PRISMA guidelines, a comprehensive literature search was conducted across five digital databases—PubMed, CINAHL, Scopus, Web of Science, and Google Scholar—to identify studies published between January 2020 and September 2025. The search yielded 332 articles; after screening, 105 full-text articles were assessed and 43 met strict PECOS eligibility criteria for inclusion. The included studies were benchmarked based on model development approaches, performance indicators, validation techniques, and consideration of patient demographic diversity. Our findings highlight substantial growth in ML-based breast cancer recurrence prediction research, yet reveal ongoing challenges such as limited external validation, inconsistent reporting, and comparatively limited use of gene-expression and imaging modalities relative to clinical data. These insights underscore the need for robust, biologically informed predictive models and motivate future work toward more interpretable and clinically actionable recurrence risk stratification tools.