Data Mining Snares Health Insurance FraudLexisNexis applies predictive modeling, a massive database, and high-performance computing cluster technology to spot health insurance fraud before claims are paid.
(click image for larger view)
Slideshow: 17 Leading EHR Vendors
As Medicare searches for ways to head off fraud
, private payers are starting to embrace predictive modeling in their own quest to stamp out insurance fraud before claims are paid. "I think the big move on the payer side is to pre-pay," according to Bill Fox, senior director of LexisNexis Health Care
, a year-and-a-half-old division of online information giant LexisNexis, a subsidiary of Reed Elsevier
. That means payers are trying to examine claims before the money goes out the door. "Virtually every big payer we talk to is thinking about it," Fox told InformationWeek Healthcare.
LexisNexis is among those joining the movement to detect fraud with advanced data mining by building analytics and risk-management capabilities into its vast data platforms. The company has built databases on 250 million people in the U.S., culled from 35 billion public records, and now is applying its analytics capabilities to health insurance.
The company analyzes its data using its supercomputer platform, which is built on top of high-performance computing cluster technology, and was made available earlier this year as an open-source platform through a new LexisNexis subsidiary called HPCC Systems. Fox says this allows for fast queries of "massive amounts of big data." The technology helps disambiguate and link data, piecing together nuggets of information to reveal collusion, both proactively and after some evidence of wrongdoing has been found.
[Legally, EHRs are double-edged swords: They protect clinicians from malpractice litigation but also put them at greater risk. See
Will Your EHR Land You In Court?]
Such analysis looks for complex patterns in the diagnosis, treatment, and billing of patient encounters that aren't easily spotted in traditional claims review.
In targeting health insurance fraud, LexisNexis looks at 15 to 18 metrics on claims and individual providers, then assigns a risk score to each healthcare provider. The system scouts for risks inherent in claims and risks inherent in each person, according to Fox, an attorney by trade who previously handled insurance fraud cases at a major law firm and has worked with the U.S. attorney's office in Philadelphia to investigate white-collar crime, including cybercrime.
For years, payers have relied on claims edits to spot errors, but they haven't been able to edit for patterns suggesting fraud because an edit focuses on a single claim and it's impossible to identify a pattern with one claim. But predictive modeling and other analytics tools can scan a series of claims to flag individual physicians and coders for extra review, Fox said, allowing payers to incorporate extra edits into future claims.
"Predictive modeling looks at outliers," Fox noted. Unusual values could indicate fraud or just simply improper coding or a physician who practices in a certain way, he said. In the past, there was no easy way of finding many errors and other unusual patterns that might merit further investigation.
Clients do tend to be payers, who are looking to stamp out waste and not be forced to pay for claims that they later learn to be improper. But Fox said that institutions such as large providers, integrated delivery networks, and accountable care organizations might be interested in this kind of service to avoid trouble with Medicare auditors and the U.S. Department of Justice as federal officials step up their anti-fraud activities.
With the advent of accountable care organizations and other elements of healthcare reform, financial risk is going to be shared among multiple entities, offering yet another reason to stamp out internal waste and fraud, according to Fox. "We'll likely see more interest from providers," he said.