What Parkland can teach other hospitals about AI in health care

Parkland Health has been helping legislators craft guidelines around the use of artificial intelligence in medicine, one of the few public hospitals doing so among private institutions like Duke and Stanford universities. Parkland’s diligence in trying to harness this new technology helps us all in Dallas County.

Hospital officials told us Parkland recently began using AI to treat trauma patients. The technology analyzes different patient statistics and compares them to past outcomes to help doctors determine whether surgery is the right option for vulnerable trauma victims or whether an operation would further jeopardize their health. The technology also updates a patient’s probability of survival in real time.

The public hospital system is also considering AI to assist doctors with paperwork. Ambient listening technologies can transcribe notes based on the doctors’ conversations with patients, giving physicians more time to focus on their bedside manner. Research has found that AI-drafted responses to text message questions from patients were perceived as more empathetic than those drafted by doctors. Still, doctors will edit notes written using AI before they are sent out.

But AI can often have biases that are hard to pin down and generate inaccurate information. Parkland leaders are trying to guard against these pitfalls by reviewing models regularly.

Parkland is a member of Duke University’s Health AI partnership corps, a group that develops standards and guidance on policy by contributing to workshops and case studies. Hospital officials said they had been contacted by a leader of the Texas Legislature’s new Artificial Intelligence Advisory Council to solicit insights about the use of the technology in health care.

Even before AI was a buzzword, Parkland officials said the hospital system was an early adopter among its public hospital peers in using electronic health records. This allowed the hospital system to collect valuable patient data and use machine learning and predictive analytics before those technologies were common. That resulted in measures of health and social factors that help doctors preempt patient needs. These indexes also help the hospital system identify neighborhoods more vulnerable to certain health conditions, such as heart disease or diabetes.

Parkland Center for Clinical Innovation, a former research department within Parkland that became a standalone nonprofit, has also harnessed predictive AI to reduce emergency-room visits and manage patient loads.

Parkland leaders say that being a public hospital actually helped them implement AI strategically. The hospital system’s budget doesn’t leave much room for experimental resources, so it uses existing AI tools that are highly vetted or collaborates with Parkland Center for Clinical Innovation on new technologies.

“We’re not looking to create AI for the sake of AI; we’re looking to find the best tool for the need,” said Brett Moran, chief medical informatics officer for Parkland.

AI is ascendant in all aspects of life and is bound to transform health care. Parkland is wise to embrace this reality and use technology in a thoughtful manner to improve treatment.

We welcome your thoughts in a letter to the editor. See the guidelines and submit your letter here. If you have problems with the form, you can submit via email at [email protected]