Detecting Fraud in Travel and Personal Accident Insurance Claims

SOMPO’s AI Fraudulent Claims Detection system reduces the manual effort involved in the claims review process and helps expedite payments

Artificial intelligence (AI) is transforming the claims review process for travel and personal accident insurance. For insurers, it has the potential to deliver “100 percent fraud detection”. For customers, it means receiving claims payments more quickly, potentially within minutes.

Fraud has long been the bane of the insurance industry, and delayed claims payments a source of customer dissatisfaction.

The two problems are related. With some customers exaggerating or misrepresenting the severity of mishaps to increase their claims, insurance companies have their work cut out combing through new claims daily to detect fraud. The manual review process is tedious and time-consuming and affects the customer experience because it holds up simple claim requests that can otherwise be processed promptly.

To unplug this bottleneck, Sompo Holdings Asia (SOMPO), a market leader in Asia’s non-life insurance industry, decided to turn to artificial intelligence (AI).

 SOMPO collaborated with AI Singapore (AISG) under the 100Experiments (100E) programme to develop a fraud detection application that could automatically flag out suspicious cases for further investigation and identify valid claims as candidates for straight-through payment.

The machine learning solution was trained to process, identify and rank suspicious travel and personal accident claims. Executed daily, the AI model gives each claim request a fraud score. Suspicious claims are then handed over to internal specialists for further investigation.

90% improvement in fraud detection

It took just seven months for the AI Fraudulent Claims Detection system to go from conceptualisation and development to pilot testing. In June 2020, the system went live at Sompo Insurance Singapore, one of SOMPO’s subsidiaries.

It has proved its mettle, taking away much of the grind involved in claims reviews whilst delivering a 90 percent improvement in fraud detection. At least 20 percent of customers also had their claims processed and payment disbursed much more quickly. In the near future, it will be possible for this to be done within minutes.

The improved workflow also enables the company to focus more on talent development. Instead of getting employees to manually review claims, they can now use their time more effectively to hone their skills in analysing and investigating suspicious cases.

These breakthroughs have not gone unnoticed by the industry. At the Singapore Business Review’s Technology Excellence Awards in June 2020, the AI Fraudulent Claims Detection system received the AI Award for General Insurance.

Author