With Organized Fraud On The Rise, Insurance Providers Require An Intelligent Fraud Detection System. Outsmart Fraud And Optimize Claims Processing For Legitimate Claims!
A leading insurance provider was facing an increased number of fraudulent auto claims. The insurer was utilizing industry-standard rule-based and network analysis to detect the fraud, but their system was insufficient—even in terms of processing legitimate claims.
Our custom-tailored system successfully blended human learning from case manager, behaviour mapping of fraudsters, and machine learning. Our innovative fraud algorithm outperformed industry-standard preexisting models by 1.4x.
The dynamic fraud flagging and alert tool helped expedite claim processing by more than 60%.
Drastic drop in fraudulent claims, led to faster processing of legitimate claims—and a rapid increase in customer satisfaction.
Insurer gained the client transparency and insights required to strengthen their approach to fraud, with data and analytics that were not previously viewed as relevant in detecting fraud.
Percentage of claims increased by 18%, leading the insurer to believe that they were being targeted by an organized fraud network. Their existing rule-based, and network analytics was able to detect about half of their suspected fraud—but with a higher than average number of false positives. This led case managers to identified independent rules to assess claims for fraud, which were not in sync with network analysis based fraud rules—and had no clear evidence of accuracy.
Dextro Analytics realised that in order to build an advanced fraud detection system, the insurer needed to be able to “think” like a fraudster—which is where Artificial Intelligence could be leveraged, and integrated with advanced analytic technology. A growing number of fraudulent auto claims are submitted by a network of allies, who work together to stage fake accidents, and claim low-impact injuries which never happened. Such claims are result of a highly organized group that often includes fraudulent activities from drivers, medical professionals, body shops, and witnesses—so we had to create firewalls that were always one step ahead.
1. Dextro’s decision engineers traced the entire journey – from policy enrolment to claim reimbursement. The detailed mapping involved mapping of doctors, lawyers, body shops, drivers, passengers, pedestrians, location, weather, geospatial data, and witnesses.
2. Our decision and data engineers then integrated first-hand information and expertise from tenured case managers, and integrated the data and insights into a custom-tailored fraud database. The end result was a superior system for identifying traits of both suspect and legitimate claims.
3. Due to multiple latent relationships among fraudsters, traditional insurance analysis could not detect multi-person collusion fraud claims. As a result, our decision and data engineers utilized a layered based graph theory to detect the relationship among all the parties involved in the fraud. This allowed the insurer to uncover incidents of organized crime.
4. Our intelligent analytic solution extracted fraud features and created a system that could flag fraud in near-real time.
1. 98% of fraudulent claims were detected through new system.
2. Our accurate fraud model also reduced the false positive by virtually 85%, providing superior service to legitimate claims.
3. Today, insurer is using our near real-time tool on hundreds of claims they receive every day.
“Dextro Analytics worked very well with our internal data science team to build highly sophisticated fraud tool. Dextro’s algorithm has enabled us to save millions of dollars in fraud every month.”
- Vice President, Risk and Fraud Analytics
Our team would love to share the nuts and bolts of the study and our best practices.
Please provide us with your contact information.