Finding Fraud Patterns in Big Data

The FBI estimates that fraud in the healthcare industry costs American taxpayers roughly $80 billion a year. Indeed, this number is likely much higher with the rise of digital technologies, the use of electronic payment systems and other modern-day factors that come into play. Opportunities to defraud the healthcare sector abound; and criminals know it.

Traditional data mining and claims analytics are simply not enough to root out the complex schemes plaguing our systems. Collusion is rampant and the only way to get ahead is to start thinking of data science, analytics and technology in new ways. Data management and linking are simply not core competencies for most healthcare organizations. The majority do not have access to the data, analytics or expertise required to uncover hidden and complex patterns leaving them exposed to fraud, waste and abuse. This is why an increasing number of healthcare organizations turn to experts that specialize in data, analytics and technology to gain crucial insights.

Claims data is the most prevalent and prominent information used for fraud detection and analysis. In our experience, we do see some plans using other sources of information like provider and member data, but not in a cohesive, integrated fashion. The types of data payers should have in contributory databases include claims, member, provider, non-healthcare data such as banking and criminal records, and other public information, which might include business-related records and relationships. Once that information is aggregated and available, the next step is to link them to form a fuller picture and identify key players, connections among the players—businesses as well as individuals—and frequently missed “patterns of behavior.”

Let’s take a fictional case of a company that produces motorized wheelchairs. A payer that notices that the company’s referrals for the wheelchairs are excessively high might discover that 70% of those referrals come from just three doctors. But further analysis of the combined data, such as criminal records, could also reveal that the company’s part-owner had been convicted of bank fraud. Because the co-owner does not submit claims, he is probably not even on the radar of the payer. But by unearthing data such as this, a payer would be better able to detect instances of criminal activity.

The ability to find the story in all of this data is immeasurable yet only possible with analytics that accurately link it all together. Linking analytics can uncover critical healthcare problems like identifying individuals associated with suspicious addresses billing a plan; uncovering social clusters around high-volume prescriptions for controlled medications; and finding complex durable medical equipment, pill mill and doctor shopping schemes.

In the era of big data, integrating linking analytics into existing payment integrity and fraud-fighting efforts should not be a question; it should be a requirement.

About The Author