Health Insurance Appeals with Generative AI - From Lab to Production

Learn how to build and deploy a generative AI system for creating health insurance appeals from denials in this conference talk from GOTO Chicago 2024. Explore the complete journey from initial model fine-tuning attempts through production deployment, covering the technical challenges and solutions encountered when developing a real-world machine learning application. Discover the iterative process of model development, including what approaches worked and failed during fine-tuning experiments, and examine the middleware architecture that integrates both in-house and external models. Understand the critical privacy protection measures implemented in the system and see the frontend implementation challenges. Follow along with a live demonstration of the appeal generation system in action, and gain insights into optimization strategies for future model iterations. The presentation addresses the practical aspects of scaling an ML system from concept to production, including model serving infrastructure, configuration management, and deployment considerations. Whether interested in the healthcare application domain or the broader technical challenges of productionizing generative AI systems, gain valuable insights into the end-to-end development process of a mission-critical AI application designed to help individuals navigate insurance claim denials.