Evaluating the Performance of Machine Learning Models for Predicting 5-Year Breast Cancer Survival: A Systematic Review and Meta-Analysis
TL;DR
Overall, the included studies demonstrated good discriminative performance, suggesting that machine learning-based models may support early identification of high-risk patients and facilitate personalised clinical decision-making.
Abstract
Simple Summary Breast cancer is a highly heterogeneous disease, making accurate prediction of 5-year survival challenging. Machine learning has emerged as a promising approach for improving prognostic accuracy by identifying complex patterns within routinely collected clinical data. This systematic review and meta-analysis evaluated the predictive performance of machine learning models developed for 5-year breast cancer survival. Overall, the included studies demonstrated good discriminative performance, suggesting that machine learning-based models may support early identification of high-risk patients and facilitate personalised clinical decision-making. Nevertheless, methodological heterogeneity, limited external validation, and variable reporting quality remain important barriers to routine clinical implementation. Further well-designed studies with standardised reporting and rigorous external validation are required before these models can be widely adopted in clinical practice.