Healthcare bias in AI: A Systematic Literature Review

The adoption of Artificial Intelligence (AI) in healthcare is transforming the field by enhancing patient care, advancing diagnostic precision, and optimizing clinical flows. Despite its promise, algorithmic bias remains a pressing challenge, raising critical concerns about fairness, equity, and the reliability of AI systems in diverse healthcare settings. This systematic literature review (SLR) investigates how bias manifests across the AI lifecycle—spanning data collection, model training, and real-world application and examines its implications for healthcare outcomes. By rigorously analyzing peer-reviewed studies based on inclusion and exclusion criteria, this review identifies the populations most impacted by bias and explores the diversity of existing mitigation strategies, fairness metrics, and ethical frameworks. Our findings reveal persistent gaps in addressing health inequities and underscore the need for targeted interventions to ensure AI systems serve as tools for equitable and ethical care. This work aims to guide future research and inform policy development, in order to prioritize both technological progress and social responsibility in healthcare AI.

人工智能（Artificial Intelligence，AI）在医疗健康领域的应用正重塑行业格局：其不仅提升了患者照护质量、优化了诊断精准性，还推动了临床工作流程的高效化。尽管人工智能在医疗领域前景广阔，但算法偏差仍是一项严峻挑战，引发了人们对不同医疗场景下AI系统公平性、公正性与可靠性的高度关切。本系统综述（Systematic Literature Review，SLR）探究了算法偏差在人工智能全生命周期——涵盖数据采集、模型训练与实际部署阶段——中的具体表现形式，并剖析了其对医疗健康结局的影响。本综述严格依据纳入与排除标准对同行评议研究开展系统性分析，明确了受算法偏差影响最显著的人群群体，并梳理了现有偏差缓解策略、公平性评估指标与伦理框架的多元路径。研究结果显示，当前在解决健康不平等问题上仍存在长期未填补的短板，本研究强调亟需实施针对性干预措施，以确保人工智能系统成为支撑公平且合乎伦理的医疗照护的工具。本研究旨在为后续相关研究提供指引，并为医疗健康人工智能领域的政策制定提供参考，从而兼顾技术进步与社会责任。