Dive Brief:
- The Coalition for Health AI released a series of governance resources this week that aim to help health systems safely roll out artificial intelligence tools.
- The playbooks, developed through community workshops and work groups that included more than 150 clinicians and health AI leaders, provide examples and guidance on implenting AI, including resources on setting up AI policies, managing third party developers and assessing risks.
- The goal is to provide a standardized, but flexible framework that health systems can use to deploy AI tools, regardless of their size or available resources, CHAI said.
Dive Insight:
Founded in 2021, CHAI is a network of more than 3,000 organizations — including health systems, technology startups and patient advocacy groups — that aims to develop guidelines on responsible AI use in healthcare.
The coalition has created a “model card” template that documents an AI model’s intended use and possible risks, as well as a registry where healthcare companies can access them. CHAI has also released responsible AI guidance developed through a partnership with the Joint Commission, as well as best practices for using AI to help determine eligibility for the safety-net insurance program Medicaid.
The latest guidance focuses on eight “critical elements” for deploying the technology at health systems, CHAI said in a press release Wednesday.
Those include playbooks on setting up an organizational structure for AI oversight — like creating a governance committee and a framework for assessing AI tools — as well as guidance on ensuring providers have the right resources in place for AI deployment, like robust cybersecurity infrastructure and tools for monitoring the model’s performance.
“We wanted to make sure the playbooks reflected the realities of the organizations that would use them,” Merage Ghane, director of responsible AI at CHAI, said in a statement. “What are the challenges, lessons learned, and resources needed to make responsible AI actionable, were just a few of the questions that guided our work.”
The coalition will continue to gather feedback on the guidance and adapt to new developments as AI governance evolves, Ghane told Healthcare Dive via email. CHAI also plans to expand the playbooks for others in the healthcare sector, like payers, she said.
The guidance comes as AI use in healthcare is quickly accelerating, including through tools to assist in medical research, document patient care or handle data analysis, according to a physician survey published by Doximity this spring. However, governance still poses a challenge for many doctors: Only 8% said the AI decision-making process at their organization was clear, and that they understood policies and guidelines.
Federal oversight is limited too. The Trump administration has largely taken a deregulatory approach to AI in a bid to spur adoption of the technology. Government officials also lambasted CHAI last year, arguing the coalition would stifle health technology startups with regulation and ultimately worsen outcomes for patients.
