Fraud prevention through prediction

Businesses are using analytics to curb losses

The Federal Bureau of Investigation estimates that healthcare fraud costs are approximately $80 billion annually. But it could be closer to $700 billion annually if fraud and improper payments are lumped together, said Julie Malida, principal of Health Care Fraud Solutions at SAS Institute Inc. This is 100 times larger than in the financial sector, she said.

[See also: Fighting fraud with predictive analytics and link analysis]

The vast amount of waste occurs mainly because there is no one group that is held accountable for stopping fraud in healthcare. Losses are absorbed by doctors, insurers, and federal and state governments.

“No one entity physically controls the reputational risk as well as dollar risk for fraud in healthcare,” Malida said. “It is so much more decentralized and lucrative than banking.”

To prevent some of the billions from being siphoned from the healthcare system, many organizations are turning toward analytics. Used across a wide range industries, analytics is the act of taking data from various sources – like medical claims or insurance records – to find errors. Some analytics are used to trace money that has been wrongfully paid and newer, predictive models are used to prevent payout before it occurs. 

The most well-know healthcare entity using analytics is the Centers for Medicare & Medicaid Services, which processes 4.8 million claims a day from providers across the country. But the usage of analytics is quickly being spread across all healthcare sectors.

[See also: UnitedHealth touts predictive modeling as solution to healthcare fraud and preventable hospitalizations]

“This is definitely the direction they had to go and the industry generally has to go there, too,” said Louis Saccoccio, executive director of the National Health Care Anti-fraud Association. “Using analytics to detect potential fraud will be absolutely necessary.”

Saccoccio said analytics break down into four major categories. The first is looking at normal rules of care and how they might be broken. For instance, using filters to see if a service was provided to one patient twice in a day or if they are providing more services than is possible in a day.

The second is looking for anomalies. Insurers can look through data to see if certain providers are claiming they consistently offer more complex or costly services than their peers.

Another type of anomaly has become more common during the recession, said Peter Millar, director of technology applications for ACL Services Ltd., an audit and risk management organization. Millar said they have seen insurance beneficiaries add family members that are not eligible to be on their health plans. For instance, if someone’s brother is out of work, a family member might claim his kids on an insurance plan. Analyzing data would show it odd for someone to have two kids one year and five the next. 

Previous
1