From Payer Policy to Structured Requirement Checklists: Leveraging Large Language Model Agents for Insurance Document Understanding

Yi-Jyun Sun, Dilek Hakkani-Tür

Keywords: LLM Agent, Document Understanding, Clinical, Prior authorization, Healthcare AI

Abstract: Prior authorization (PA) remains a primary driver of healthcare burnout, yet applying Large Language Model (LLM) agents to formal policy interpretation remains largely unexplored. Currently, manual synthesis of requirements from unstructured documents is error-prone and frequently delays patient care. We propose an \textbf{agentic architecture} for \textbf{Insurance Document Understanding} to model the transition from dense clinical prose to structured, CPT-indexed requirement frameworks. We evaluate this approach by quantifying \textbf{extraction accuracy} against human-annotated policies. By demonstrating high fidelity in semantic decomposition, this work establishes a novel baseline for the automated interpretation of clinical administrative discourse.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 164