为高风险医疗问答设计可选择性认证，让系统智能判断每条结论是否可信。
Claim-Selective Certification for High-Risk Medical Retrieval-Augmented Generation
- 将回答拆解为可验证的命题，由意图感知的筛选器分类为全信、部分信等四类
- 在测试集上实现0%未支持命题风险，92%动作准确率，97%精确率
- 适合对医疗AI可靠性要求极高的临床决策辅助场景
高风险医疗问答中的检索增强生成系统通常采用单一答或弃权决策方式评估，但不同命题可能有混合证据支持、需附加条件或相互矛盾。本文提出命题选择性认证：将每个回复分解为可验证命题，根据检索到的证据评分，并通过意图感知筛选器映射至‘完全支持’、‘部分支持’、‘冲突’或‘弃权’四类。在主要弱标签证书协议下，开发集（n=314）上达到UCCR=0.0000（未支持命题风险），PAU=1.0000，PAU精度=0.9901，动作准确率=0.9204；测试集（n=319）上对应值为UCCR=0.0000，PAU=0.9967，PAU精度=0.9739，动作准确率=0.8997。UCCR衡量证书定义内的未支持命题风险，源缺失反事实切片用于评估空证据下的弃权表现。快捷控制量化了由源信息和意图元数据解释的动作标签先验，源/证据新奇切片刻画了跨域迁移边界。最终界面实现了在混合证据下，动作预测与证据关联命题选择的分离。
原文摘要 · Abstract (English)
Medical RAG systems in high-risk QA settings are often evaluated through a single answer-or-abstain decision, but mixed evidence may support one claim, require conditions for another, and contradict a third. We study claim-selective certification: each response is decomposed into verifiable claims, scored against retrieved evidence, and mapped by an intent-aware selector to {full, partial, conflict, abstain}. On the primary weak-label certificate protocol, whose real-source-only dev/test rows cover the naturally occurring non-abstain actions, the full system records UCCR=0.0000, PAU=1.0000, PAU Precision=0.9901, and action accuracy=0.9204 on dev (n=314), and UCCR=0.0000, PAU=0.9967, PAU Precision=0.9739, and action accuracy=0.8997 on test (n=319). UCCR measures unsupported-claim risk within the certificate definition, and a source-missing counterfactual slice evaluates abstain under empty evidence. Shortcut controls quantify the action-label prior explained by source and intent metadata, while source/evidence-novel slices characterize transfer boundaries. The resulting interface separates action-label prediction from evidence-linked claim selection under mixed evidence.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。