This research-in-progress report presents a framework that combines large language models (LLMs), small language models (SLMs), Failure Mode and Effects Analysis (FMEA), and the ISO 31000 risk governance standard to detect and manage hidden risks in U.S. healthcare payer–provider contracts. Healthcare payer–provider contracts frequently contain ambiguous reimbursement language, prior authorization conditions, and appeal timelines that create financial and compliance exposure but remain difficult to detect before disputes arise. The framework identifies the intent behind ambiguous contract language—not just the clause classification. These intent interpretations are translated into structured risk signals, scored using FMEA’s Risk Priority Number model, and mapped to ISO 31000 lifecycle stages to drive governance decisions. A proof-of-concept pilot on representative clause scenarios yields early evidence of feasibility. The framework aims to help contract lifecycle management (CLM) teams shift from reactive dispute resolution to proactive risk management. Specific calibration parameters will be reported in the complete dissertation. This report outlines the research problem, the approach, preliminary observations, and reflections on challenges encountered so far.

Keywords: healthcare contracts, FMEA, intent-aware AI, ISO 31000, risk governance, contract lifecycle management, large language models