The Use of AI in Claims Data

A Compendium of Artificial Intelligence Use in Utilization Review

from AppriseMD

Payers deny claims at a rate that climbs higher each year. However, the increasing use of artificial intelligence (AI) to automatically deny claims without review creates a new issue for hospitals and revenue cycle managers struggling to reduce denials. While healthcare insurance companies promote the use of AI as a denial mitigation tool to process claims more efficiently, three class action lawsuits filed late last year are raising concerns that insurance companies use AI to deny claims without ever examining patient records. As the use of AI in healthcare claims management widens, the importance of careful oversight is needed. Any use of automation for claim denials needs to be scrutinized by providers to ensure the denial is appropriate, which also satisfies numerous state insurance laws requiring proper evaluation of healthcare claim denials. This publication is a review of publicly available news articles discussing the use of AI in claim denials.

HOW ARE PAYERS USING AI?

Class action lawsuits filed against UnitedHealthcare (UHC), Humana and Cigna in 2023 claim misuse of AI in denying medical necessity claims. One suit alleges Humana and UnitedHealthcare “wrongfully deny Medicare Advantage (MA) members’ medical claims”¹ using an AI tool owned by UnitedHealth Group (UHG). A separate Cigna case emerged surrounding the use of a procedure-to-diagnosis algorithm, referred to as PxDx. The lawsuit filed against Cigna alleges the algorithm denied claims in batches, with 300,000 claims being denied in two months’ time in 2022.² These cases hold significance because of the sheer number of patients covered by these companies. UHC and Humana together account for 47% of all MA enrollees or 14.4 million patients, with UHC covering 29% of these patients. Cigna accounts for a much smaller portion – 2% of all MA enrollees.³ 

UHC & HUMANA

The charges against UHC and Humana, filed in November 2023, allege denials were based on “rigid and unrealistic predictions for patients’ recoveries.”⁴ The lawsuits contend that Medicare Advantage patients were denied medical coverage for post-acute care through use of an AI tool without a review of a patient’s case and unique medical situation. The suits state that patients often begin receiving payment denials after only two weeks of post-acute care. That is in contrast to Traditional Medicare’s timeline, which allows up to 100 days of post-acute care following a qualifying hospital stay.

Both UHC and Humana have been accused of using the automated data tool, nH Predict, to override a physician’s clinical determination regarding a patient’s post-acute care. The lawsuit alleges nH Predict is a “deeply flawed AI algorithm” that estimates how much post-acute care a patient will need based on medical case data from millions of patient records. Ryan Clarkson, the founder of the law firm representing the plaintiff in the suit against UHC, said “This is an example of how AI is being utilized not to help people but to line the pockets of corporations and their shareholders.”⁵ Despite an error rate estimated to be close to 90%⁶, UHC continues to use this product and plans to defend itself against these allegations.

CIGNA

Earlier in 2023, ProPublica reported Cigna appeared to be in violation of state laws by denying claims in large batches using an algorithm which by-passes individual physician review, a requirement in many states⁷. A lawsuit filed against Cigna accused the payer of denying coverage to patients through its use of PxDx which enabled the company’s medical directors to deny large batches of claims without reviewing the patients’ medical records.

THE USE OF AI IN HEALTHCARE CLAIMS

Technology supported decision-making drives the use of AI in healthcare. As clinicians spend more time documenting in the electronic medical record (EMR), AI can be an asset when it comes to analyzing the data quickly and efficiently. A scan of the headlines at the end of 2023 shows hospitals are embracing AI as a revenue cycle tool as well. A Revcycle Intelligence article from Dec. 27 stated that, “74% of hospitals use some revenue cycle automation,”⁸ while a HealthLeaders year-end piece asked, “How will revenue cycle leaders expand AI use in 2024?”⁹ But as hospitals use AI to “streamline processes and automate tasks,” payers will increasingly do the same. Oversight is needed on both sides. The White House began pushing for more oversight in late 2023 with the December release of “Delivering on the Promise of AI to Improve Health Outcomes,” which outlines “voluntary commitments on the safe, secure, and trustworthy use and purchase and use of AI in healthcare.”¹⁰ The document outlines the principles health systems, payers and healthcare companies agreed to as it pertains to the use of AI in healthcare outcomes. The document asks healthcare entities to “align the AI industry with the FAVES’ principles, emphasizing that AI in healthcare should lead to outcomes that are fair, appropriate, valid, effective, and safe.”¹⁰ Twenty-eight providers and payers acknowledged their commitment according to the White House release.

THE USE OF AI IN MEDICINE

Prior to the White House document release, the American Medical Association (AMA) weighed in on the discussion of AI in healthcare. The AMA House of Delegates utilizes the term “augment intelligence” as a “conceptualization of artificial intelligence that focuses on augmented intelligence’s assistive role, emphasizing that its design enhances human intelligence rather than replaces it.”¹¹ AMA emphasizes that “As the number of augmented intelligence-enabled health care tools and systems continue to grow, these technologies must be designed, developed and deployed in a manner that is ethical, equitable, responsible and transparent.”¹² AMA urges caution as physicians engage with augmented intelligence-enabled technologies in the healthcare setting, especially if the developer is not transparent in providing appropriate disclosures such as whether the AI tool “impacts patient care, access to care, medical decision making, communications, or the medical record.” The organizations are further calling for government oversight to mitigate the risks of AI, specifically stating that “voluntary agreements or voluntary compliance is not sufficient.”

THE REGULATION OF AI IN HEALTHCARE

With major payers in the spotlight for using AI to deny claims, and allegations that the use of AI has overridden the physician’s judgment, legislation regulating the future of AI in healthcare may be coming. Meanwhile, insurance companies could decide to follow the suggestion of one former executive and use AI to be a first round of automatic denials. “Why not just deny them all and see which ones come back on appeal? From a cost perspective, it makes sense,” one former executive stated.⁶ Payers have a process to appeal denied claims, but hospitals should continually evaluate denial management processes to meet the rising tide of denials. Insurance companies are likely to take the gamble that the number of appeals won’t match the denials, and they will come out ahead financially.

SUMMARY

If the allegations in the lawsuits filed against the major payers are accurate, it is clear that the use of AI for denial management is not aligned with a patient’s best interest. Further, if the AI-generated denials from the nH Predict algorithm were reversed 90% of the time when appealed during an Administrative Law Judge (ALJ) hearing as communicated by Clarkson⁵, then patients need to appeal these denials. As payers continue to use AI for claim management, hospitals should stay vigilant and employ a consistent approach to appealing denied claims. Greater denial management efforts are needed to improve a hospital’s revenue cycle; providers need to scrutinize automated denials to ensure appropriateness.

This review is a compendium of publicly available news articles discussing the use of AI in claim denials. AppriseMD believes in holding payers accountable with physician-driven solutions dedicated to resolving reimbursement and reinforcing revenue integrity for hospitals and health systems. AppriseMD will continue to monitor the use of AI, enhancing denial management and physician advisor solutions to supplement automated processes aimed at optimizing hospital reimbursement.

SOURCES

1. Emerson, Jakob, “Humana used AI tool from UnitedHealth to deny Medicare Advantage claims, lawsuit alleges,” Becker’s Healthcare. December 13, 2023.
2. Minemyer, Paige, “Cigna hit in August with another lawsuit over claims denials through PxDx,” Fierce Healthcare. September 22, 2023.
3. Ochieng, Nancy; Biniek Jeannie Fuglesten; Freed, Meredith; Damico, Anthony; and Neuman, Tricia, “Medicare Advantage in 2023: Enrollment Update and Key Trends,” Kaiser Family Foundation. August 9, 2023.
4. Watkins, Morgan, “Lawsuit claims Humana uses AI to deny necessary health care services to Medicare Advantage patients,” Louisville Public Media. December 14, 2023.
5. Pierson, Brendan, “Lawsuit claims UnitedHealthcare AI wrongfully denies elderly extended care,” Reuters. November 14, 2023
6. Payerchin, Richard, “UnitedHealthcare used AI to deny patients’ health insurance coverage, lawsuit says,” Medical Economics. November 16, 2023.
7. Rucker, Patrick; Miller, Maya; and Armstrong, David, “How Cigna saves millions by having its doctors reject claims without reading them,” ProPublica. March 5, 2023.
8. LaPointe, Jacqueline, “74% of Hospitals Use Some Revenue Cycle Automation,” Revcycle Intelligence. December 27, 2023.
9. Ray, Jasmyne, “How will revenue cycle leaders expand AI use in 2024?,” HealthLeaders. December 28, 2023.
10. Brainard, Lael; Tanden, Neera; and Prabhakar, Arati, “Delivering on the promise of AI to improve health outcomes,” The White House. December 14, 2023. Diaz, Naomi, “Health systems commit to Biden’s AI principles,” Becker’s Healthcare. December 14, 2023.
11. “Augmented intelligence in medicine,” American Medical Association. November 28, 2023.
12. “Principles for Augmented Intelligence Development, Deployment, and Use,” American Medical Association. November 14, 2023.