A new report from the Pittsburgh-based Center for Connected Medicine and KLAS Research finds that a majority of health care systems and hospitals have yet to develop detailed policies surrounding the use of artificial intelligence.
The survey found only 16% of those who responded to the survey had put into place a systemwide governance policy for the use of AI and access to data. That’s likely due to the fact that the AI technology — especially those that are using the relatively new ChatGPT and Bard — are still early in development and adoption, said Dr. Rob Bart, UPMC’s chief medical information officer. UPMC is a founding member of the Center for Connected Medicine.
“It wasn’t overly surprising to me that the percentage of health care systems and policies governing systemwide policies governing artificial intelligence is relatively low,” Bart said. “I suspect over the course of calendar year 2024, that number is going to change dramatically.”
UPMC has been on the leading edge of AI uses in health care but also, just as importantly, those policies and data privacy. Bart said that UPMC put into place a comprehensive policing guiding the usage of AI last year. Only five of the 31 health care executives surveyed were in the same position.
“Our timing was good relative to where the industry is and what we needed to govern,” Bart said.
UPMC’s policy takes into account ethical considerations, health systems’ regulatory obligations and also making sure that AI use conforms with its standards.
The health care industry has been an early adopter of some types of artificial intelligence and machine learning, primarily in radiology where AI helps read patterns in patient imaging.
Survey respondents said they believed generative AI, such as ChatGPT and Bard, would help physicians make clinical decisions based on data and analysis, streamline the patient record-keeping process, automate other processes and help doctors communicate with their patients.
Bart, in addition to his job as chief medical information officer, is also a practicing pediatric intensivist. He believes that generative AI could be used in the provider-patient interactions as a documentation tool that can record information automatically during a visit and then input information into a document doctors complete later.
That could ease the physician workload, allowing more time with patients. But Bart said it will also allow for less time with a computer screen during the patient visit and more face-to-face time.
“What I like about what it does is it starts to make the technology retreat in the room, even though the technology is key to the operation,” Bart said.
Executives told the survey that they are actively working on ways to employ AI across its operations, with 11 of 23 saying their strategies include showing staff and managers the proven benefits. Others are providing training and increasing exposure to AI, as well as identifying the proper use cases and making it easier to use.
But Bart doesn’t think, even with extensive use of AI, that it will ever replace the clinician directly in front of the patient.
“It’s not going to replace the physician,” Bart said. “It’s more of an enabler.”
