Health Systems Prioritize Artificial Intelligence Governance, Oversight

February 15, 2024 - Health systems are increasingly prioritizing the development of artificial intelligence (AI) oversight efforts as they continue to navigate the potential promise and pitfalls of these tools in healthcare, according to a report published today and shared with HealthITAnalytics by the Center for Connected Medicine (CCM) at the University of Pittsburgh Medical Center (UPMC).

The report, “How Health Systems are Navigating the Complexities of AI,” surveyed executives from almost three dozen health systems on their approaches to overseeing the technology.

As AI rapidly advances, healthcare organizations continue to investigate how these tools can help automate administrative tasks and reduce clinical documentation burdens, among other use cases. However, this increased interest and swift development makes the pros and cons of AI in healthcare challenging to balance, as stakeholders have raised a plethora of concerns around data privacy, clinician over-reliance, patient trust, and more.

To overcome these hurdles, healthcare organizations are developing strategies for AI oversight, like the National Academy of Medicine’s AI Code of Conduct. CCM’s report emphasizes that health systems themselves are taking this approach as well, with 16 percent of respondents indicating that their organizations have a system-wide AI governance policy in place.

Many more noted that their health systems have formed governance committees consisting of senior leadership to oversee AI deployment.

These findings highlight a shift in how healthcare organizations are looking at these technologies. While the report demonstrates that few health systems have formal, written policies around the use of AI, and fewer have specific policies for newer generative AI tools, organizations are seriously considering how the technologies could help them meet strategic goals.

“There are many ways health care can and will benefit from AI, including freeing up our clinicians to focus more on caring for patients and helping systems more efficiently process a range of tasks,” said Robert Bart, MD, chief medical information officer for UPMC, in a press release. “But it is essential that health care executives also take seriously the responsibility to protect our patients’ privacy and health data.”

The report also found that executives are interested in generative AI tools, particularly how they can be integrated into existing tools like electronic health record (EHR) systems. Roughly 70 percent of executives reported that they plan to or have adopted AI solutions via EHR vendors.

Respondents further indicated that they expect generative AI to enhance healthcare by improving efficiency, automating repetitive tasks, and bringing more visibility to clinical decisions.

These insights underscore the importance of oversight and governance for the successful implementation of generative AI tools.

“Before adopting generative AI technologies in health care, it’s crucial for executives to clearly define their objectives and establish measurable benchmarks,” said Jeffrey Jones, senior vice president of product development at UPMC Enterprises, the innovation, commercialization, and venture capital arm of UPMC. “Regular evaluations are essential to adjust strategies as necessary. Generative AI is not a one-time fix, but a dynamic tool that requires attention and calibration.”

To support health systems as they navigate the complexities of AI, national healthcare organizations have been working to provide guidance, education, and best practices around the use of these tools.

Last week, the American Health Information Management Association (AHIMA) launched its AI Resource Hub, which is designed to provide healthcare and health information (HI) stakeholders with knowledge about non-clinical AI tools.

The resource is based on findings from AHIMA’s “Artificial Intelligence Tools for Documentation and other Non-Clinical Work in Healthcare” white paper, a survey of over 200 hospitals and 1,000 clinics across the United States exploring how these tools are deployed and the challenges they present.