New guardrails for Medicare Advantage plans' use of AI may not be clear enough, experts told members of the Senate Finance Committee.
CMS sent a message to Medicare Advantage plans Feb. 6, clarifying how new prior authorization rules set forth by the agency apply to AI. The agency wrote that AI programs can be used to assist in coverage determinations, but it is the plans' responsibility to "ensure that the algorithm or artificial intelligence complies with all applicable rules for how coverage determinations by MA organizations are made."
The agency will also up its auditing of denials in Medicare Advantage plans, according to the message.
At the hearing, Michelle Mello, PhD, professor of health policy at Stanford University, told the committee CMS should implement more specific guidelines on requirements for meaningful human review of claims denied by algorithms.
"Audits by CMS need to look very closely, as I believe they intend to, at denials where algorithms were involved," Dr. Mello said.
UnitedHealthcare and Humana, the largest Medicare Advantage insurers, are facing lawsuits alleging they wrongfully denied members care using an AI-powered algorithm. A spokesperson for Optum, the UnitedHealth Group subsidiary that operates the algorithm, told Becker's the tool is not used to make coverage determinations.
A spokesperson for Humana told Becker's its augmented intelligence maintains a "human in the loop" whenever AI is used, and "adverse coverage decisions are only made by physician medical directors."
Ziad Obermeyer, MD, Blue Cross of California Distinguished Associate Professor of Health Policy and Management at the University of California Berkeley, told the committee AI learns from historical data and can reinforce existing trends rather than improve them.
"[AI] trolls through millions of records, and sees there are some privileged people with great insurance who stay in nursing homes longer than they should, and there are also vulnerable, underinsured people who are often kicked out too early," Dr. Obermeyer said. "Rather than undoing that problem, the AI reinforces it and encodes it as policy."
Well-designed AI programs could make these decisions better than humans, Dr. Obermeyer said.
"It could look at the patient's X-ray, it could look at the public transportation in their neighborhood, it could look at the layout of their house, and integrate all those things into a far better judgment than a doctor is able to make on who needs to be in that nursing home and who doesn't," he said.
Sen. Elizabeth Warren, a frequent critic of Medicare Advantage, said CMS should prevent MA plans from using AI in prior authorization until it can confirm algorithms do not result in wrongful denials of care.
"It takes the bad information and accelerates it, or [accelerates] the information that is bad practice," Ms. Warren said.