代理治理与对抗性验证：面向策略约束的LLM医疗申诉生成
Agentic Governance and Adversarial Verification for Policy-Constrained LLM Healthcare Appeal Generation
- Sliced Health
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
针对医疗申诉生成中单代理架构的不足，提出多代理AGVF框架，通过约束马尔可夫决策过程与引文接地门控，实现策略约束下的可靠生成，并在千例合成案例中验证零违规。
AI中文摘要：
拒赔管理每年给美国医疗系统造成约2600亿美元的管理性行政开销。大型语言模型（LLM）与检索增强生成（RAG）能够生成流畅的临床文本，但单代理架构在高风险医疗场景中表现不佳：它们会引入缺乏支持的临床细节，并丢失层级化支付方策略的逻辑结构。我们提出AGVF（代理治理与对抗性验证框架），一种在明确策略与证据约束下生成医疗必要性申诉的多代理架构。AGVF将申诉综合建模为基于五个代理的约束马尔可夫决策过程（CMDP）：策略形式化、证据检索、差距分析、对抗性批判与门控综合。我们证明，在固定策略约束图上的迭代优化能够单调地减少证据缺陷，并以完整的满足性前沿或局部证据缺口终止。一个确定性的引文接地门控可防止无充分证据的主张进入共享状态。我们提供了参考实现，并在基于去标识化公共医院出院数据参数化的1,000个合成申诉案例上进行了验证。验证确认所有AGVF案例均实现零引文接地违规，且每个回合中缺陷均单调减少；移除该门控后违规率升至100%，证实其关键作用。本研究未使用真实患者记录，也未测量临床疗效。因此，AGVF为医疗领域策略约束的LLM生成提供了一种有理论支撑的代理架构和经过验证的参考实现。
英文摘要：
Claim denial management costs U.S. healthcare approximately $260 billion annually in administrative overhead. Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) can produce fluent clinical text, but single-agent architectures fail in high-stakes healthcare: they introduce unsupported clinical details and lose the logical structure of hierarchical payer policy. We propose AGVF (Agentic Governance and Adversarial Verification Framework), a multi-agent architecture for medical-necessity appeal generation under explicit policy and evidence constraints. AGVF models appeal synthesis as a Constrained Markov Decision Process (CMDP) over five agents: policy formalization, evidence retrieval, gap analysis, adversarial critique, and gated synthesis. We prove that refinement over a fixed policy constraint graph monotonically reduces evidence-deficiency and terminates with either a complete satisfying frontier or a localized evidence gap. A deterministic citation- grounding gate prevents assertions without admissible evidence from entering shared state. We provide a reference implementation and validate it on 1,000 synthetic appeal cases parameterized from de-identified public hospital discharge data. The validation confirms zero citation-grounding violations across all AGVF cases and monotone deficiency reduction in every episode; ablating the gate raises violations to 100%, confirming it is load-bearing. The study uses no real patient records and does not measure clinical efficacy. AGVF thus contributes a theory-backed agentic architecture and verified reference implementation for policy-constrained LLM generation in healthcare.
