5/1/2009

Techno-Sleuthing

New technology, better data and improved modeling can finger insurance fraudsters

By Richard Della Rocca

It’s difficult to overstate the importance of battling insurance fraud. Even under normal economic conditions, tens of billions of dollars are lost every year through payment of fraudulent and inflated insurance claims. In today’s recessionary climate, the impact of fraud-related losses rises considerably. Coupled with continuing pressures on insurer investment portfolios, this unnecessary outflow of cash is detrimental to the industry’s fiscal health.

Claims fraud takes many forms and infiltrates every line of business. The problem is not limited to experienced criminals who may work singly or in tandem with crooked physicians, phony treatment centers, and hired “claimants.” With job losses and foreclosures surging, even individuals with no prior criminal history are falsifying insurance information—from padding actual claims to reporting home break-ins that never occurred—to obtain cash.

Collecting and Analyzing Data Is Key
Effective fraud detection depends upon the ability to collect and analyze vast amounts of data and identify activities and patterns that are indicative of potential fraud. For years, the industry has submitted claims information to the ISO ClaimSearch database, which currently holds more than 600 million claim records across all lines of insurance.

A critical first step in determining whether a claim is meritorious or suspicious is to review the individual’s claims history. When insurers input claims into the ISO ClaimSearch system, they automatically receive a report that indicates if any of the fields—from name to Social Security number to address—match prior claims in the system. Adjusters will examine those matches closely, looking for patterns of claims with similar participants, injuries, loss descriptions, or targeted businesses. With further inquiry, insurers can investigate to see if there are red flags that could point to fraud. Examples include the claimant who uses several post office boxes, the chiropractor who is used repeatedly in slip-and-fall incidents, and the Social Security number that is associated with more than one name.

But well-grounded suspicions are just the starting point. Investigating and building a case is a much more complicated task. The good news is that technological developments are strengthening insurers’ ability to analyze data more thoroughly and manage investigations more efficiently.

Link Analysis and Data Visualization
Link analysis and data visualization software tools look for links between data elements, such as all claims linked to the same VIN number. Link analysis software can search through millions of pieces of data and connect the element to every other data element with which it has a relationship. Through a combination of such powerful sorting technology and data visualization software that generates a graphic display of the data connections, investigators can quickly see direct and indirect relationships among multiple pieces of data, such as parties linked to multiple addresses, telephone numbers, vehicles, and claims. These tools can reveal connections within company data, data from external sources, or a combination of sources, providing claims investigators insight and knowledge to make decisions about which claims to investigate further. Such systems also can create presentation-quality graphics that support law enforcement activities and criminal prosecution where warranted.

Predictive Modeling
Predictive modeling systems analyze current and historical data to make predictions about the future. This technology has been used for years in the financial services industry to generate credit scores. Increasingly, predictive modeling is being applied to insurance to predict the probability that a particular claim is suspicious. One example is a scenario-based model that assigns a numerical ranking to a claim based on the likelihood of fraud. The system compares the attributes of a current claim against proven fraud indicators, using the same rules that are applied by experienced claims professionals when they manually evaluate claims. The model scores the claim to reveal the likelihood of fraud. This score provides adjusters with a quick and easy way to distinguish between suspicious and meritorious claims.

Other analytic techniques are being applied to the insurance fraud problem. These include statistically based predictive models that analyze characteristics of known fraudulent claims to identify good predictors of fraud that can be applied to new, incoming claims; neural network technology to uncover emerging fraud patterns; and text mining applications that can analyze the vast amount of claims information contained in documents and notes to find additional clues of fraudulent behavior.

By drawing upon all of these advanced analysis tools, insurance investigators can supplement tips, referrals and their own suspicions. These tools compress weeks of complex analysis into minutes, reducing the time needed for analysis and allowing investigators to work more productively.

Strides in Data and Processing
Even the most advanced claims analysis capability is dependent upon the quality of the data being analyzed.In recent years, an industry fraud-data working group composed of senior claims executives and representatives from ISO and the National Insurance Crime Bureau (NICB) developed plans to improve industry resources in the fight against fraud by improving the data reporting process to ISO ClaimSearch. As a result of this joint effort, the group implemented several fraud data initiatives aimed at making ISO ClaimSearch even more actionable for fraud detection and investigations.

While developments in data collection and analytic methods are significantly improving insurers’ ability to identify potential fraud, additional technological improvements can improve claims workflow. Such advances are making managing claims investigations faster, easier and, ultimately, more effective.

For example, claims fraud investigations require careful planning, coordination, analysis and documentation. Case management software for special investigative units (SIU) can provide investigators with a computerized system to organize notes and case files. This allows SIUs to better handle insurance investigations and provides at-a-glance status reports to streamline many of the processes for investigations. These applications can track daily activity, to-do lists, time, mileage and expenses. By handling many administrative functions, these systems give investigators more time for analysis. This software also can allow SIU managers to monitor investigation activity and measure the performance and productivity of their operations.

ISO ClaimSearch also offers an integrated Web portal to access external and public records information. Claims handlers and investigators can easily verify Social Security numbers or addresses, find unlisted phone numbers, and check whether a physician has been sanctioned. As they research claims, claims staff can request additional information, including event data recorder information, medical records, and motor vehicle reports. The ability to access this information directly, rather than repeatedly leaving the system to search for information on the Internet, results in a huge cumulative time savings in the course of a claims handler’s day. In addition, insurers can report questionable claims to the NICB and state fraud bureaus directly through the system.

Technology Working for Insurers in Numerous Ways
Insurers’ ability to fight insurance fraud has improved dramatically in recent years from advancements in data sharing and technology. The field of data analysis has been revolutionized by link analysis, data visualization, predictive modeling, and other analytics. The ISO ClaimSearch database has been enhanced to collect better-quality data, a process that continues to be bolstered by the industrywide conversion to Universal Format. Process improvements, such as one-click access to public records and case management software, help streamline claims investigations. Other process improvements—integrated suspicious claims referrals and regulatory reporting, for example—free up more time for analysis. All of these continually evolving technologies work in concert to help insurers detect and prosecute fraud and, ultimately, to improve their bottom line.
Richard Della Rocca is vice president of ISO Claims Solutions.



Rich Della Rocca is president of ISO Claims Analytics. He can be reached at rdellarocca@verisk.com.

Also By This Author

Top Industry News

Powered by : Business Insurance


SEA