Recommended guidelines concerning artificial intelligence and health equity: a scoping review

The use of artificial intelligence (AI) technologies in healthcare has potential to improve patient care, expand access, and optimize efficiency. Yet, a growing body of evidence warns of health risks for marginalized and minoritized populations as a result of algorithmic bias.

To synthesize leading scholars' recommendations for equitable AI implementation in healthcare settings and health services research, we conducted a scoping review of current literature.

We identified and analyzed 166 relevant, peer-reviewed articles. Most articles (70%) originated in the United States. Only 13% provided a specific definition of "health equity." Included manuscripts generally emphasized that ethical human oversight must accompany AI-use. The engagement of diverse groups of stakeholders during model development, implementation, and evaluation is also paramount. Patient and provider education on AI's capabilities is also critical for building trust and equitable delivery of care.

To enhance the field's ability to implement AI in ways that address health disparities and inequities, future research needs to amplify discipline-specific expertise and more global perspectives. By explicitly prioritizing health equity, transparency, inclusion, and education, there is potential to meaningfully regulate AI in healthcare to redress systemic inequities in care and advance health and well-being in populations throughout the world.