Integrating artificial intelligence in medicine can come in many forms.
One algorithm could be made to help spot signs of cancer in scans earlier than they can be seen by the human eye, aiding in earlier diagnosis for things like breast cancer.
Other WRAL Top Stories
Another concept could involve a system which transcribes clinical notes during a doctor’s appointment to reduce the workload for medical staff.
Duke University officials hosted a forum on Thursday with local and national health and technology experts to discuss responsible use of AI.
A panel discussion, led by Duke Health Chief Data Scientist Michael Pencina, involved Duke Health’s director of evaluation and governance, UNC Health System chief medical informatics officer, Microsoft’s global chief medical officer and leaders with the U.S. Department of Health and Human Services.
HHS Deputy Secretary Andrea Palm said implementing AI in healthcare in its many forms will require strict and clear guidelines for both companies that create the technology and the hospitals which use them.
“We are at the beginning of this journey, and so the opportunity to really engage with developers with different types of settings in which this technology will be deployed is really critical to thinking about how we get into that level of detail and make sure we get it right," she said.
Following the forum, Duke led tours through Duke University Hospital to showcase a way AI could enhance patient care.
Operating rooms are currently equipped with cameras spaced out on walls around the OR to record a procedure from multiple angles.
“They capture image, voice, video and also some of the other data feeds that happen through anesthesia,” said Dr. Ozanan Meireles with Duke Health.
After a surgery, the current system requires someone to go through hours worth of footage so then a surgical team can discuss what went right or what could be improved.
Meireles explained if the FDA were to approve AI for use in surgery, the technology could potentially be integrated into the existing camera system to speed up the time it takes to review footage.
He also added algorithms could be made that could potentially monitor real-time data of a surgery to notify surgeons of possible problems with a patient on the table before they occur.
Echoing tales of caution from other health officials in the forum, Meireles stated this new technology would first need to be thoroughly tested in diverse settings.
He added he hoped one day the technology could then be scaled for use beyond the Duke Health system to help with hospital operations across the globe.
“For example, if you’re going to drive a car and you don’t have a drivers license, you don’t understand what’s happening on the road,” explained Meireles. “I would say as those systems become more sophisticated, I think people should be educated and eventually have a license to utilize those systems.”
Meireles said ultimately he expects clinicians and surgeons to make final medical decisions “based off the recommendation of AI models.”
The exact guidelines that companies and health systems would need to follow in order to use AI in medicine remain unclear.
Health experts add those regulations could look different for when the tech is used for something like a procedure versus a routine doctor’s visit.