Payers and providers are locked in an “AI arms race” over coding and claims, but some industry leaders are acknowledging the battle may be a distraction from providing patient care.
“The bots battling the bots is a real thing,” Sachin Jain, MD, SCAN Group president and CEO, told Becker’s.
“We’re spending all this effort trying to figure out how to get paid and how to deny payment, instead of figuring out how to streamline the actual experience of care so that people who need help can get it,” he said. “We should all take a pause and say, ‘Can we end this madness?’ Because it is absolute madness.”
Centene and Elevance Health, among other insurers, have accused providers this year of “aggressive coding,” with AI-assisted clinical documentation tools heightening that scrutiny. One health system’s clinicians reported reduced rates of burnout while using ambient AI, but also thousands more in revenue per clinician.
While payers escalate their AI resources to authenticate claims and streamline prior authorization requests, some providers say insurers are outpacing them and creating an imbalance.
“We in the provider sector are chasing this a bit,” Dennis Laraway, executive vice president and CFO of Cleveland Clinic, said.
“We don’t control the purse strings, so we’re often responding to claims edits, rules, bulletins and things of that nature that guide, from the purse-string holders — from government payers, CMS, Medicaid plans to small, medium and large commercial plans across our communities,” he said. “We do our best to adopt change, create billing editors and claim editors, and try to align with the rules and with changes over time to minimize as much friction as we can.”
Cleveland Clinic still sees more than 15% of claims originally denied despite its attempts to follow payer rules, with that figure going down to below 2% after the appeals process.
“But all of the time, energy and resources consumed to get it from 15% to sub-2% is unsustainable,” Mr. Laraway said. “That’s arguably also unnecessary and unsustainable for the payers and the plans. If they’re watching the same report card that we are — an initial 15% ends up being sub-2% — that doesn’t seem to be productive friction in healthcare for either side of the table.”
One payer wanted to clarify: AI is not working alone.
“We leverage AI to identify whether a claim could potentially be a fraudulent claim,” Angle Health CTO and co-founder Anirban Gangopadhyay said. “We add notes onto a claim, but we never process anything that’s flagged without a human reviewing it. It goes back to that human-in-the-loop paradigm.”
Just as providers have been under pressure with their AI use, so have payers. A recent lawsuit over coverage denials by UnitedHealth Group directly addresses mainstream concerns about AI overriding physician judgment. Families of two deceased Medicare Advantage members challenged the company’s use of a specific algorithm, alleging premature denials of necessary services. New Hampshire lawmakers even recently proposed legislation to cut back insurers’ use of AI for claims audits, as well, highlighting state unease.
Mr. Gangopadhyay said, as a growing health plan, his team has not witnessed “flagrant cases of fraud” from providers — usually just coding issues that providers fix when prompted. But larger plans are noticing a trend.
In a third-quarter earnings call with investors, UnitedHealthcare CEO Tim Noel noted how ongoing elevated costs are at least partially “related to more service intensity per encounter being billed by health systems and by providers.”
“Some of the ways that’s coming through are higher cost sites of service where lower costs are available,” he said. He added that the insurer is seeing more services attached to hospital visits, more specialists rounding during inpatient stays and “a bias” toward higher diagnosis-related groups.
As insurers and providers point fingers, a lesson is emerging. Without rethinking the system as a whole, Dr. Jain said the industry is losing sight of the point of healthcare. Both payers and providers have a role to play in the balancing act of how AI could reinforce current tensions versus an industry-wide move toward better patient outcomes.
