Health insurers are beginning to adopt AI to support prior authorization decisions. But is this a good thing? Experts weighed in during a panel discussion held by KFF on Thursday.
One panelist said she has some questions about the use of AI in prior authorization, and added that there should be more transparency on this topic and how often prior authorization requests using AI are overturned.
“I think that as we see the use of AI increase, one question we have is, what’s the data that’s going into these algorithms? What data are these algorithms based on? Are they the most current data? Do these algorithms include old studies that may not reflect the best medical knowledge that we have right now? How often are they being updated? Are they being encouraged to deny care, at least at the first level?” said Anna Schwamlein Howard, principal of policy development at the American Cancer Society Cancer Action Network.
However, there has recently been more scrutiny of the use of AI in healthcare, which is a good thing, according to Kaye Pestaina, vice president and director of the Program on Patient and Consumer Protection at KFF. She noted that just last week, there was a Senate hearing on AI in healthcare.
Another panelist echoed the need for transparency when it comes to AI in prior authorization. However, he noted that AI and newer technologies also have the opportunity to improve and speed up the prior authorization process.
“We talk about cancer, it takes over four weeks to get in to see an oncologist or radiation oncologist today, and I hate to think that part of that delay is the result of people having to deal with prior authorization. So any decrease in the latency period of getting people treated is an important thing. And I think as long as you’ve got the transparency and you can understand what these algorithms are doing, then I think it’s potentially a very important improvement overall in the process. I wouldn’t be afraid of it,” said Dr. Troyen Brennan, adjunct professor of health policy and management at Harvard T.H. Chan School of Public Health. Brennan is also a former executive at CVS Caremark and Aetna.
Dr. Fumiko Chino, a radiation oncologist at Memorial Sloan Kettering Cancer Center, said that she welcomes “our new computer overlords with some caveats.”
“We know that datasets are very flawed and that for example, marginalized populations are much more likely to have undocumented stage or they may be missing key elements from their EMR notes that would lead to barriers and therefore may disproportionately face denials,” Chino said. “Then you’ve trained a machine based on a dataset that is essentially racist. I think that’s ultimately what we have to fight against.”
How will the use of AI in prior authorization affect patient trust? Schwamlein Howard noted that the average patient isn’t thinking about this.
“They’re focusing on getting better,” she said.
Photo: Piotrekswat, Getty Images
