Can Big Data Nix Fraud?

While insurance fraud has steadily bedeviled the industry since its inception, the tools to address the problem have improved markedly in recent years. Now insurers can sift through burgeoning data sets and employ increasingly sophisticated analytics to catch wrongdoers. Insurance Networking News asked Stuart Rose, global insurance marketing manager for Cary, N.C., analytics provider SAS, about how insurers can leverage the latest technologies in fraud detection.

INN: What will the increasingly large volumes of data that insurers manage mean for their ability to detect fraud?

SR: It could be argued that the more data available the easier it is to hide fraud, however the converse is usually true. The larger the amount of data available, the greater the likelihood of discovering fraud, especially when insurers go beyond viewing each claim in isolation and view claims at an enterprise level. Today, insurers are beginning to supplement their own claims information with third party data, such as National Insurance Crime Bureau (NICB) alerts, ISO claims history, even social media data, to increase the possibilities of discovering suspicious activities. Hence with this increasing volume of data, insurers need to implement data management and analytical software that enables them to not only detect suspicious behavior but to prevent future similar fraudulent activities.

INN: What new analytic capabilities can help insurers detect fraud?

SR: At most insurers, the technology that is currently in place to support fraud fighting is a mixture of business rules and database searches. While these techniques have proven successful in detecting known fraud patterns, today insurers need to invest in new analytical capabilities to detect unknown and complex fraud activities. These analytical capabilities include: anomaly detection, predictive modeling, text mining and social network analysis.

* Anomaly detection uncovers new fraud scams by finding those elements that vary from normal. Key performance indicators associated with tasks or events are baselined and thresholds set. When a threshold for a particular measure is exceeded, then the event is reported. Outliers or anomalies could indicate a new or previously unknown pattern of fraud.

* Predictive models use past fraud events to produce fraud-propensity scores. Adjusters simply enter data, and claims are automatically scored for their likelihood to be fraudulent and made available for review. Using predictive modeling makes it possible to reduce false positives and capture more fraud.

* Text mining enables insurers to analyze unstructured data such as police reports, medical records, even e-mails. Considering that 80 percent of claims data is unstructured, it is imperative to use text mining to analyze this information for scripted words or multiple claimants using identical phrases.

* Social network visualization tools let investigators actually see network connections so they can uncover previously unknown relationships and conduct more effective and efficient investigations.

Also, today's analytical fraud technology contains the ability to "learn" from experience to get better at fraud detection and pattern identification. This learning characteristic enables the software to adapt and increase in sophistication as more and more intelligence is gathered over time. The more analytical the tools, the more chance of detecting fraud in the early stages and predicting potential areas of fraud before the criminals have even uncovered the opportunity. Automation also places less reliance on the human element and provides greater accuracy and fewer false positives to chase down.

INN: What new types of fraud should insurers be on the lookout for?

SR: The objective of fraudsters is to devise new ways of fraudulently obtaining money from an insurance company that is undetected. Hence predicting new types of fraud is like predicting the lottery numbers. Fundamentally, it is still the same old frauds being committed-it's just in slightly new ways. One example is the emergence of online application fraud and ghost broking. Online application fraud is similar to rate evasion, the customer or insured alters some information to obtain lower rates, such as "accidently" transposing the year of birth from 1987 to 1978. "Ghost" brokers work by acting as a "middle man" between the customer and other brokers/insurers and could provide the customer with a fraudulent insurance policy either by insuring incorrect information, or giving the insured a fake insurance policy.

Organized fraud activity, although not emerging, is rapidly increasing, especially with medical providers such as excessive treatment, inflated billing and solicitation. This type of fraud is particularly prevalent in "no-fault" states such as Florida for PIP claims.

Finally, watch for auto glass fraud, as insurers introduce fast track claims processing to settle these claims without adjuster evaluation, the amount of fraud according to the NICB has increased by more than 400%.

For reprint and licensing requests for this article, click here.
Analytics Security risk Data security Policy adminstration Data and information management
MORE FROM DIGITAL INSURANCE