The campaign against “waste, fraud and abuse” in the federal government has become a staple of political campaigns. Both those who monitor for fraud after the fact and those who try to prevent taxpayers from getting swindled in the first place are using Big Data and new analytical tools to spot the scammers. However, experts on a panel from a range of federal agencies and in the private sector also cautioned that it’s important to remember the human element of fraud detection amid the embrace of new technologies.
During a panel discussion last month at MeriTalk’s Second Annual Big Data Brainstorm in Washington, the experts noted that Big Data and other uses of analytics can help them spot patterns and focus their attention and resources.
Using Big Data and Analytics to Spot Patterns
Tony Trenkle, chief health information officer for IBM’s Global Healthcare Industry unit, says that at the Centers for Medicare and Medicaid Services, recent estimates indicate that 5 to 10 percent of all Medicare spending is fraudulent, and that within Medicare the figure is about 6 percent.
Trenkle, who formerly worked at CMS as well as at the Social Security Administration, says agencies can use technology to “edit” or process claims before they are paid to stop fraudulent payments from being distributed. He also says agencies can look at patterns and develop models that can be used to see when anomalies occur. “It’s an iterative process,” he says. “It works. And over the years a number of agencies have come a long way with it.”
Mike Mashburn, director of the Data Analytics Lab in the Office of the Inspector General of the U.S. Postal Service, says that by looking at cases of events that actually happened — such as when a postal worker slipped and fell on the job and claimed disability payments — fraud investigators can use that data to find patterns.
“If we now apply what we see going backwards from convicted cases to data that is moving forward you can begin to see a pattern,” he says. Then, that information can be sent to investigators to see if fraud is being committed.
Issues with the Data That’s Coming In
While using Big Data tools can be useful in finding patterns and predicting fraud, a potential inherent flaw is the quality of the data that agencies’ systems and tools are ingesting, the experts say.
Johan Bos-Beijer, director of analytics services for the General Services Administration, says that data ambiguity and disparity are actually “fraud enablers.” Sometimes, he says, people who are managing a program are actually conducting the fraud, and they can take advantage of the fact that data is not often shared within an agency. “That’s an open door and enabler to fraudsters,” he says.
Bos-Beijer says he encourages those he is training to spot fraud to think like scammers, and put on their “amateur theatrical cap” to think about how the data could be used to create a scam — and then reverse-engineer the fraud.
Remembering the Human Element
Trenkle says it’s important to remember the human factor in rooting out fraud. CMS has a program called Senior Medicare Patrol, in which volunteers are trained to work with beneficiaries. They help identify anomalies on Medicare summary notices and statements to look for fraud and prevent identify theft, he says.
Bos-Beijer says that sometimes federal agency workers will buy a piece of software and think that will intuitively solve problems with fraud, or think that by merging data sets they will get the answers they are looking for.
“And 90 percent of the time, it’s the human that has the gut instinct, that has the behavioral reference, that understands what another human being might do to take advantage of a program,” he says. “I think that’s very critical and [it is important] not to lose sight of that too. The technology definitely enhances that capability, and certainly ramps up in a large program the capability of identifying and rapidly targeting and preventing fraud.”