How Do Artificial Intelligence Applications in Breast Cancer Impact Health Equity, Particularly in Diagnosis, Treatment, and Outcomes of Diverse Groups: A Critical Literature Review (Preprint)

BACKGROUND The integration of Artificial Intelligence (AI) in breast cancer care has the potential to impact health equity, yet the extent of this impact remains underexplored. Understanding how AI applications influence health disparities, including racial, ethnic, and socio-economic factors, is crucial for developing effective and equitable interventions. OBJECTIVE This review aims to provide a comprehensive analysis of current evidence on the impact of AI technologies on health equity in breast cancer (BC) care. It focuses on how AI influences disparities in diagnosis, treatment, and outcomes among different racial, ethnic, and socio-economic groups, and critically assesses the quality and robustness of studies evaluating AI interventions intended to address these inequities. METHODS Adhering to PRISMA 2020 guidelines for Systematic Reviews, this Critical Literature Review analyzed studies published between January 2019 and August 2024. A comprehensive literature search across Medline, Embase, Web of Science, and Scopus was conducted to identify peer-reviewed research addressing AI applications in breast cancer and their impact on health equity. Studies were evaluated for their focus on health equity outcomes, and methodological rigor. The quality of each study was assessed using the Public Health Critical Appraisal Checklist (PHCA) and the Ruling Out Bias Using Standard Tools in Machine Learning (ROBUST-ML) checklist. RESULTS The review included seven studies. The findings of the study highlighted three main themes of AI's role in reducing healthcare inequalities: forecasting biological and socio-economic factors affecting outcomes, enhancing decision-making fairness, and use of AI to reduce healthcare inequalities in a global context. CONCLUSIONS The studies highlight AI’s potential in addressing health inequalities in breast cancer through its capacity to analyze complex datasets, incorporate race/ethnicity-specific models, and function in resource-limited settings. However, its impact depends heavily on data quality, dataset representativeness, and the ability to mitigate inherent model biases. Therefore, policymakers should focus on improving dataset inclusivity, establishing benchmark datasets, standardizing data collection, and addressing data poverty in low- and middle-income countries. Future researchers should ensure dataset representativeness and generalizability, conduct prospective or randomized controlled trials to assess real-world applications, and carefully select predictor variables through multidisciplinary collaboration. Adherence to standard reporting guidelines is also essential to enhance reproducibility and ensure the reliability of AI applications in breast cancer.