Unlocking efficiency: Transformed claims processing with generative AI (MAM228)
Unlocking Efficiency: Transforming Claims Processing with Generative AI
Introduction
Doug Armstrong, Head of Data Science at Securian Financial, and Prithvi Mazumder, Technology Fellow at Deloitte, present a collaborative effort to improve claims processing using generative AI.
The goal is to share the story of how they identified, understood, and solved the problem, as well as the technology considerations and benefits of the solution.
The Problem: Manual Claims Adjudication
Claims adjudication is a highly manual process, particularly for a product like hospital indemnity insurance.
The process involves several steps:
Claimant submits claim and uploads documents
Examiner reviews documents to capture key data points
Examiner compares data to rules and approves/rejects claim
Decision is communicated, and payment is issued
Data Challenges
Data is captured in various random formats (e.g., tilted, rotated images)
Data context varies across different hospital/provider documents
Documents can be in different languages
Inconsistent scanning and conversion to images
Technology Selection and Implementation
Collaborated with Deloitte and internal experts to identify core capabilities needed:
Ingestion
Text extraction
Insight derivation using large language models
Applying rules and making recommendations
Leveraged AWS services to build a secure, scalable solution.
Focused on maintaining a "human in the loop" approach, with the AI providing summary information to the claims adjudicator.
Designed the solution to be repeatable and reusable across the enterprise.
Benefits and Lessons Learned
Reduced document completeness check time from several minutes to under 5 minutes.
Improved customer satisfaction, with claims adjudicators now "disappointed" when they don't see the AI-generated summary.
Enabled potential for triage and automation of simple claims in the future.
Importance of collaboration, communication, and including all stakeholders in the process.
Significance of maintaining a "human in the loop" approach, with the AI supporting and enhancing the claims adjudicator's work.
Demo Simulation
Demonstrated a simulated claims adjudication process, highlighting the integration of the claimant statement and hospital discharge summary, automated data extraction, and application of exclusion rules.
Showcased a feature that allows the claims adjudicator to view similar historical claims and their resolutions, helping to train new adjudicators.
Next Steps and Conclusion
Continued exploration of opportunities to further automate and optimize the claims processing workflow.
Emphasis on maintaining the secure and privacy-focused foundation for all data and technology initiatives.
Overall, a successful collaboration and a compelling story of leveraging generative AI to transform a manual, labor-intensive process.
These cookies are used to collect information about how you interact with this website and allow us to remember you. We use this information to improve and customize your browsing experience, as well as for analytics.
If you decline, your information won’t be tracked when you visit this website. A single cookie will be used in your browser to remember your preference.