Claims Fraud Detection

3AI August 15, 2020

Leading Global P&C Insurance Company

Problem Statement

• Customers’ current SIU was geared towards proactively identifying and detecting claimant related frauds.
• Fraud related to medical provider were proving to be difficult to identify and work upon
• Customer had to rely on referrals from regulatory and industry bodies to identify suspect cases and wanted to explore ways and means to pro-actively identify provider fraud through statistical means

Analytics Led Approach

• Framework on claims fraud detection was developed to identify the potential losses due to fraud or suspect claims and thereby helped to  reduce claims leakage to a great extent
• Given are high level process steps which were followed
  • Sensitive data encryption
  • Load the data from multiple sources
  • Data cleansing
  • Data transformation
  • Business Significant variables deduction
  • Distribution analysis
  • Correlation analysis
  • Multi- dimensional Analysis Fraud analysis through data profiling

Business Impact

• An analytical solution that focuses on providers rather than transactions, and looks for patterns in provider behavior was developed
• The solution methodology was end-to-end: ingesting medical invoice data, creating its own data assets and classifying outlier behavior using unsupervised machine learning techniques
• Solution outcomes were tangible and allow self-serve configuration of business rules.
• Recommendations can readily be evaluated by follow-up investigation and $ business benefits are easy to estimate and establish

Critical Success Factors

• Key suspect indicators and patterns were identified which indicated possible provider frauds
• Identified $ 6.4 million was charged for medical bills by suspect medical Providers