Report: Most health systems lacking an AI policy

Nearly two-thirds (65%) of health system executives in a recent survey reported having no policy in place to address the use of artificial intelligence at their organizations despite AI’s rapid inroads into health care. 

Of the roughly one-third (35%) that do have a policy, fewer than half have an AI-specific policy. The rest have broader IT and data use policies that include AI but aren’t specific to it. The low numbers are likely because many organizations are still in the early stages of evaluating and implementing AI solutions, according to researchers who conducted the study. Others are waiting for federal regulations to be issued before creating their own policies. 

“How Health Systems Are Navigating the Complexities of AI” is the title of the 24-page report, which was commissioned by the University of Pittsburgh Medical Center’s (UPMC) Center for Connected Medicine. The survey, conducted during October and November 2023, received responses from 35 hospital and health care executives of whom just under half (46%) were presidents or CEOs. 

When it came to having an AI policy, even fewer (11%) respondents reported having a policy specific to the use of generative AI. Yet, it is in generative AI where respondents envision some of the most improved efficiencies coming from. Automating routine tasks and streamlining the updating of patients’ electronic health records (EHR) are examples. Both would remove a significant time burden from clinicians long frustrated by the need for manual documentation, which would free them up to focus more directly on patient care. 

More than 70% of respondents are counting on their EHR vendors to do the heavy lifting of integrating AI solutions into patient record-keeping systems for them, in part because it would preempt the need for dealing with additional vendors. Many large, established vendors such as Microsoft and Epic are also now touting AI in their products and have HIPAA-compliant protocols already embedded in the software. Only 10% of respondents do not anticipate implementing AI solutions directly from their EHR vendors. 

More broadly, respondents envision using AI to improve other operational tasks such as scheduling, patient flow and work queue management. Identifying how you plan to implement AI to improve efficiencies and effectiveness at your health care organization should be done in tandem with establishing policies and guardrails around its use, according to the report’s authors.  

“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 commercial 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.” 

Nearly half (48%) of respondents to the survey reported working at health organizations with more than 500 beds. A further breakdown showed 14% having between 500 and 1,000 beds and 34% with more than 1,000 beds.