Unmasking racial bias in medical AI: a narrative review of evidence and implications

Abstract

Artificial Intelligence (AI) is transforming healthcare, promising improve-ments in diagnosis, treatment, and patient outcomes. However, racial bias persists and feeds inequities. This review inspects how bias manifests in medical AI do-mains, identifying unrepresentative training data and proxy variables. It explores mitigation strategies and knowledge gaps, spotting the interdisciplinary approach to fortify equitable and accountable medical AI.