Implementation in organizations

Tools for different aspects of AI implementation in health systems.

Guiding Principles to Address the Impact of Algorithm Bias on Racial and Ethnic Disparities in Health and Health Care

Chin MH, Afsar-Manesh N, Bierman AS, et al. JAMA Netw Open. 2023 Dec;6(12):e2345050.

This article outlines guiding principles to address racial and ethnic bias in healthcare algorithms. Drawing on expert input, it proposes a framework centered on equity, transparency, community engagement, fairness evaluation, and accountability across an algorithm’s life cycle. As clinical algorithms become more widely used, clear standards are needed to prevent them from reinforcing existing health disparities and ensure they promote equitable care.

Clinical use of artificial intelligence requires AI-capable organizations

Novak LL, Russell RG, Garvey K, Patel M, Craig KJT, Snowdon J, Miller B. JAMIA Open. 2023 Jul;6(2):ooad028.

This article argues that using artificial intelligence in clinical care safely and effectively requires health systems to become AI capable. Rather than focusing only on the technology, the authors emphasize building governance structures, workforce training, documentation standards, and evaluation processes to manage bias, risk, and implementation challenges. Emphasis is on ensuring health care institutions are structurally prepared to use it responsibly and equitably rather than simply adopting AI.

Enhancing trust in AI through industry self-governance

Roski J, Maier EJ, Vigilante K, Kane EA, Matheny ME. J Am Med Inform Assoc. 2021 Jul;28(7):1582–1590.

The article argues that sustaining trust in healthcare artificial intelligence requires more than technical improvements and regulatory rules. It calls for stronger industry self-governance where developers, users, and stakeholders collaboratively define and uphold standards for AI risk mitigation and ethical practices. It highlights the risk of declining trust if high expectations for AI aren’t met by trustworthy performance and oversight. The authors describe how voluntary certification and accreditation processes could complement regulation to promote responsible AI development and use.