A new Senate bill would require Medicare to test two tools routinely used by credit card companies to prevent fraud: Artificial intelligence (AI)-trained algorithms to detect suspicious activity and a system to quickly alert Medicare patients on whose behalf payment is being sought.
Senator Mike Braun (R-IN) recently introduced the Medicare Transaction Fraud Prevention Act, which calls for a 2-year test of this approach.
The experiment, targeted to start in 2025, would focus on durable medical equipment and clinical diagnostic laboratory tests and cover Medicare beneficiaries who receive electronic notices about claims.
The legislation would direct the Center for Medicare and Medicaid Services (CMS) to test the use of predictive risk-scoring algorithms in finding fraud. The program would be modeled on the systems that credit card companies already use. Transactions could be scored from 1 (least risky) to 99 (most risky).
CMS would then check directly by email or phone call with selected Medicare enrollees about transactions considered to present a high risk for fraud.
Many consumers have benefited from this approach when used to check for fraud on their credit cards, Braun noted during a November hearing of the Senate Special Committee on Aging. Credit card companies often can intervene before a fraudulent transaction is cleared.
“There’s no reason we wouldn’t want to minimally at least mimic that,” Braun said at the hearing.
Asking Medicare enrollees to verify certain purchases could give CMS increased access to vital predictive data, test proof of concept, and save hundreds of millions of dollars, Braun said.
Concerns Raised
So far, Braun has only one cosponsor for the bill, Senator Bill Cassidy, MD (R-LA), and the bill has drawn some criticism.
Brett Meeks, executive director of the Health Innovation Alliance, a trade group representing technology companies, insurers, and consumer organizations, objected to requiring Medicare enrollees to verify flagged orders. CMS should internally root out fraud through technology, not burden seniors, Meeks told this news organization.
Meeks said he has been following the discussion about the use of AI in addressing Medicare fraud. Had a bill broadly targeted Medicare fraud through AI, his alliance might have backed it, he said. But the current proposed legislation has a narrower focus.
Focusing on durable medical equipment, for example, could have unintended consequences like denying power wheelchairs to people with debilitating conditions like multiple sclerosis, Meeks said.
But Braun’s bill won a quick nod of approval from a researcher who studies the use of AI to detect Medicare fraud. Taghi M. Khoshgoftaar, PhD, director of the Data Mining and Machine Learning Lab at Florida Atlantic University, Boca Raton, Florida, said he sees an advantage to Braun’s approach of involving Medicare enrollees in the protection of their benefits.
The bill does not authorize funding for the pilot project, and it’s unclear what it would cost.
Detecting Medicare Fraud
The federal government has stepped up Medicare fraud investigations in recent years, and more doctors are getting caught.
A study published in 2018 examined cases of physicians excluded from Medicare using data from the US Office of Inspector General (OIG) at the Department of Health and Human Services.
The OIG has the right to exclude clinicians from Medicare for fraud or other reasons. Chen and coauthors looked at Medicare physician exclusions from 2007 to 2017. They found that exclusions due to fraud increased an estimated 14% per year on average from a base level of 139 exclusions in 2007.
In 2019, CMS sought feedback on new ways to use AI to detect fraud. In a public request for information, the agency said Medicare scrutinizes fewer claims for payment than commercial insurers do.
About 99.7% of Medicare fee-for-service claims are processed and paid within 17 days without any medical review, CMS said at the time.
A version of this article appeared on Medscape.com .