Multiple commercial data sources bolster fraud investigations

Applying pattern recognition software to multiple databases offers an efficient and frequently effective way for investigators to root our potential fraud, according to the New York Times.

Although data analytics doesn't necessarily expose instances of outright fraud, it does identify anomalies that can point investigators in the right direction. In some instances, the data is nuanced. In one case, investigators from the New York Human Resources Administration (NYHRA) tracked a Bronx woman suspected for fraud based on $50,000 in benefits she and her three children received from Medicaid, according to the article.

Using other databases, including business ownership records, car registration and property ownership, investigators found the family was hiding an electrical contracting business and three residential properties. This month, the Bronx district attorney charged the woman with grand larceny and making false statements to a public office.

The NYHRA told the Times that this new tactic has allowed staff members to complete nearly 30,000 investigations and uncover $46.5 million in fraud in 2014, compared to 48,000 fraud investigations worth $29 million in 2009.

Other states, such as North Carolina, also use predictive data analytics provided by IBM, SAS and LexisNexis to compare state benefit payments with other databases, such as luxury car purchases, which can help point investigators in the right direction.

"The data-mining process is extremely important," Steven Banks, commissioner of the NYHRA, told the Times. "It allows us to zero in on likely fraud so we don't divert resources to finding what otherwise might be a needle in a haystack."

Others have their doubts. Jay Stanley, senior policy analyst for the American Civil Liberties Union, told the Times the use of commercial data raises questions about whether fraud accusations are fair and accurate. Data scientists counter that they are careful to validate data against more than one database when building a case.

Health fraud experts say predictive data analytics has been, and remains, crucial in quickly identifying and stopping fraudulent payments. The federal government's Fraud Prevention System (FPS) has reportedly saved the Medicare program $54.2 million in actual and projected savings in just two years.

Furthermore, the old "pay-and-chase" model that the industry previously relied on is becoming obsolete with the emergence of predictive software, which has led to an increase in public-private partnerships designed to root out fraud.

For more:
- read The New York Times article