Home Insurance Fraud Detection

Overview

• In recent years, there has been a rise in problematic operators exploiting large-scale natural disasters to solicit housing repairs and claims processing on behalf of disaster victims, charging exorbitant fees. To address this, Aioi Nissay Dowa Insurance established a Fire Insurance Support Center in April 2022, staffed by skilled employees to handle cases with a high likelihood of problematic operator involvement. However, these operators' tactics have become increasingly complex and sophisticated, necessitating a strengthened damage assessment framework. 
  To tackle this, the company collaborated with AIOI R&D LAB and Mind Foundry, an AI venture from Oxford University, to develop an AI system that scores the likelihood of problematic operator involvement on a case-by-case basis using 1.1 million past claims data. This system, combined with expert measures, aims to further protect customers and ensure appropriate claim payments.
   

Detail

Background

• Problematic operators exploiting natural disasters not only charge exorbitant fees but also engage in deceptive tactics, such as presenting ineligible damages as disaster-related or deliberately damaging property, attempting to implicate policyholders in insurance fraud. 

• To protect customers and combat fraudulent claims, urgent action was required.
   

Hypothesis

• The company possesses vast past insurance claim data and expertise in handling problematic operators. AIOI R&D LAB-OXFORD has an established fast-paced AI development framework. Mind Foundry has successfully developed multiple fraud detection models. 

• Therefore, by collaborating, the three parties can develop a fair, transparent, and effective solution in a short timeframe. 

 

R&D

• Leveraging 1.1 million past claims data (contract details, repair estimates, problematic operator trends) along with historical conversation data, the model enables keyword-based and key-phrase analysis. 

• It scores the likelihood of problematic operator involvement for each case and quantitatively provides its reasoning. 

• Close collaboration between Aioi Nissay Dowa Insurance’s claims investigation team, AIOI R&D LAB-OXFORD, and Mind Foundry ensures the AI model is trustworthy, effective, and thoughtfully designed.
   

Next step

• Integrate the AI into the investigation decision-making process by prioritizing high-score cases, creating an efficient and effective claims investigation framework. 

• Skilled employees will investigate high-score cases, improving the precision of results. The investigation outcomes will be used to further train the AI system, enabling it to adapt to new fraudulent tactics through continuous algorithm updates.