About
Rule and AI/ML( Hybrid) based indicator to identify suspected claim at FNOL and various claim stages
Innovation presentation
Fraudulent claims are a huge loss for the insurance business. TATA AIG believes in just and fair settlement of claims in a cost-efficient manner by a proactive and time-bound approach but also curb any kind of unethical practice. However, until recently past, suspicious claim investigation was manually managed through process and training. With rapid expansion and growth, there was a need to shift from this subjective approach to a more objective granular and AI/ML approach augmented by our domain expertise. This solution uses historical data to build an AI-based system supplemented with human experience (claims team) to score individual claim transactions.
The solution has benefited the claims team by automating the workforce management basis of the score which resulted in an approximately 30% lift in repudiation due to fraudulent reasons.
Uniqueness of the project
All the zonal heads were invited to give their feedback and their critique was positively evaluated which further helped in refining the system.
There is a workflow that is driven by risk categories (SENSITIVE/HIGH/MEDIUM/LOW/NIL) and all the claims that are flagged as HIGH/MEDIUM are scrutinized by the senior-most heads accordingly.
The Data Science team is working on a Machine Learning model parallel that will eventually replace the rule engine as rules are biased on the human experience.