Data-mining information helps MetLife detect fraud. (Technology: Technology Notes).

Data mining, a way of using technology to uncover fraud rings, has helped MetLife Auto & Home link data electronically at a significantly faster pace than possible with human effort alone according to its special investigation unit.

MetLife Auto & Home began providing the information gleaned from data mining to field investigators fighting fraud in 2001, said John Sargent, the company's corporate special investigation unit manager, at the 2003 Insurance Fraud Management Conference in Baltimore. The conference is hosted by the National Insurance Crime Bureau and Insurance Services Office Inc.

Although the quality of data was poor at the start, Sargent realized he needed help with the software from an information technology expert at his company, who mapped fields, determined what could fit into servers and quickly improved the system, he said.

Now, link-analysis software helps the company collect information in half a day, rather than the 100 man hours it would have taken otherwise, said SIU analyst James Murphy.

MetLife Auto & Home, a unit of MetLife Inc., uses data mining to react to data coming in to the SIU and to actively search for fraud, he said. The company uses its own claims data from the past two years.

For instance, the company will look for policy-holders committing rate evasion, such as lying about where they live or where they garage their car to pay lower premiums, by checking their ZIP codes with their home telephone numbers to see whether the cities match, Murphy said. In one case, MetLife learned of a rate-evasion ring in New York and Massachusetts and searched and compared policy states and home area codes. After just 2 1/2 hours of fraud investigation, the company found 107 claims linked to the ring, he said.