This week, the Coalition for Health AI (CHAI) introduced a new set of governance tools aimed at assisting healthcare organizations in the safe implementation of artificial intelligence (AI) technologies. These resources, crafted through extensive workshops and collaborative efforts involving over 150 health professionals and AI experts, include practical guidance on establishing AI policies, managing third-party developers, and evaluating associated risks.
The initiative’s primary objective is to equip health systems with a flexible yet standardized framework for integrating AI tools, suitable for various sizes and resource levels, as noted by CHAI.
Launched in 2021, the Coalition for Health AI consists of a diverse network of more than 3,000 entities, encompassing health systems, tech startups, and patient advocacy organizations, all focused on promoting responsible AI applications in the healthcare sector.
Among its contributions, CHAI has developed a “model card” template, which outlines the intended purposes and potential risks of AI models, along with a registry that allows healthcare organizations to access these documents. Additionally, in collaboration with the Joint Commission, CHAI has published guidelines on responsible AI usage and best practices to assist in determining eligibility for Medicaid, the safety-net insurance program.
The coalition's latest guidance outlines eight essential elements for the effective deployment of AI technologies within healthcare systems. These resources include playbooks designed for establishing an organizational structure for AI governance—such as forming a dedicated oversight committee and developing a framework for assessing various AI tools. The guidance also emphasizes the necessity for healthcare providers to have adequate support systems for AI implementation, which includes strong cybersecurity measures and performance monitoring capabilities.
Merage Ghane, the director of responsible AI at CHAI, expressed that the design of the playbooks was influenced by the real-world challenges faced by their users, including the lessons learned and resources required to ensure that responsible AI practices are effectively put into action.
Ongoing feedback collection will allow the coalition to refine its guidance continually and adapt to the evolving landscape of AI governance, as Ghane communicated in an email to Healthcare Dive. Plans are also in place to broaden the playbooks' applicability to various stakeholders within the healthcare ecosystem, including payers.
The release of these guidelines comes at a time when the use of AI in healthcare is rapidly expanding, with applications ranging from aiding medical research to streamlining patient care documentation and data analysis, according to a recent physician survey conducted by Doximity. However, many healthcare professionals still face governance challenges; only 8% reported clarity in their organization’s AI decision-making processes and an understanding of relevant policies and guidelines.
Moreover, the current federal oversight on AI remains limited. The previous Trump administration adopted a deregulatory stance intended to facilitate the technology's adoption, but this approach has drawn criticism. Last year, government officials accused CHAI of potentially hindering health tech startups with excessive regulation, which they argued could ultimately lead to poorer outcomes for patients.
