Follow Datanami:
September 11, 2016

Predictive Analytics Helping Insurers Spot Fraudulent Claims

Suresh Aswani

As the digital age presents new security risks for the financial services industry, spotting fraudulent activity continues to be an increasingly complex and expensive issue for insurance companies.

According to the Insurance Information Institute, property/casualty fraud costs the U.S. insurance industry up to $32 billion each year. That means, in the time it takes to read this sentence, approximately $8,000 has been paid out in fraudulent claims.

Difficult economic conditions have caused a surge in suspicious activity over the past few years, driving up expenses for insurers, threatening profitability, and forcing companies to compensate by charging exorbitant policy premiums. The Federal Bureau of Investigation estimates that insurance fraud costs the average U.S. family an extra $400-$700 per year in increased premiums.

Fraudsters are continuously finding new methods of deceiving insurance companies, so much so that it is virtually impossible for manual processes to keep pace with malicious activity or recognize falsified claims before they are paid.

In this rapidly evolving marketplace, insurance companies are striving to quickly adapt their anti-fraud efforts and data management capabilities in order to reduce loss ratios. Advanced analytics and modeling tools are allowing insurers to sift through growing data volumes, delivering business insights which help them spot and investigate falsified claims faster and with more accuracy.

Predictive modeling employs a rules-based engine to quickly process vast quantities of data and uncover suspicious patterns. By aggregating structured and unstructured data from historical databases, claims management systems, and even third party sources like social media, analytics allows insurers to compare the characteristics of new claims against those of past losses in order to flag potentially fraudulent activity.

Predictive analytics can bolster anti-fraud efforts through a number of applications:

  • Popular fraud methods can be identified in newly-opened claims and new fraudulent techniques pinpointed in real time.
  • Text mining solutions can be leveraged to collect streaming data, which assists claim investigators (e.g. data collected from a claimant’s social media page may be analyzed to determine the viability of their claim details, such as the extent of their injuries).
  • Scoring, which predicts the likelihood of fraud, can be delivered early in the claims intake process, so specialists can ask more probing questions or immediately involve investigators to confirm fraudulent activity.
  • Claims processing can be optimized, also helping to improve the accuracy and efficiency of underwriting, subrogation, and recovery.

For today’s insurers, survival is dependent on transforming outdated IT infrastructures to include the right mix of computational speed and storage to power Big Data analytics. By adopting a scalable, high-performance Big Data platform that delivers insights in real time, insurers can enhance anti-fraud measures, better forecast risk, and considerably reduce losses from fraudulent claims.

Hybrid computing models are enabling companies to eliminate information siloes and renew existing legacy systems at reduced costs. These on-premise solutions will effectively leverage data points throughout the claims process, harnessing up-to-the-minute data to support predictive modeling and generate future trends.

For the insurance industry, profit is inextricably linked to risk reduction and prevention, and data-driven insights are key to detecting new methods of fraud. Companies that can scale to accommodate immense data growth will be able to quickly identify fraudulent claims, enhance the efficiency of investigations, and ensure business longevity and profitability.

Datanami