What if a denied prior authorization request could be appealed using ChatGPT? It's an idea being discussed at Grady Health in Atlanta.
"It is possible to take a payer denial letter, load that and payer manuals into a large language model, and use that to draft a cogent and compelling appeal letter using the payer's own policies," Murry Ford, vice president of revenue cycle at Grady, told Becker's.
Mr. Ford was discussing the health system's prior authorization strategy during a virtual RCM event in June.
He said Grady has been putting more resources into the prior authorization process, leading to a decrease in denial rates. The system is not piloting any ChatGPT-based process that involves personal health information or claims data, but it is "playing around with taking publicly available payer manuals" to build searchable libraries of payer policies.
"Our hope is that we can load in denial letters, then bounce the claim off the payer guidance and the letter to see if everything makes sense before you draft an appeal," Mr. Ford said. "I think the future is being able to take payer policies and using a large language model like ChatGPT to manage appeals."
Other health systems have recently deployed large language models for various administrative and clinical workflows.
In July, The New York Times reported that physicians are increasingly turning to generative AI to appeal denied prior authorizations. Some experts told the outlet the process could become an "arms race," where insurers will use AI technology to deny requests.
"Their AI will deny our AI, and we'll go back and forth," Robert Wachter, MD, chair of the department of medicine at the University of California, San Francisco, told the Times.