TRACE：面向肿瘤学大语言模型的可部署树关系结构增强框架
TRACE: Deployable Tree-Relational Structure Enhancement for Oncology LLMs
浏览论文内容
中文总结 AI 辅助
TRACE通过可更新的树关系结构增强肿瘤学大语言模型，分离离线学习与在线推理，在零样本下提升分类与问答性能，优于RAG和GraphRAG，并支持可解释的临床推理证据路径。
中文摘要 AI 辅助
大语言模型在肿瘤学应用中的使用日益增多，但其预测往往缺乏对明确医学结构的强有力支撑。我们提出TRACE，一个面向肿瘤学大语言模型的可部署树关系增强框架。TRACE将昂贵的离线结构学习与轻量级在线推理分离：肿瘤学概念和关系被组织成可更新的树关系结构，利用基于语言模型损失推导的证据进行细化，并在推理时作为紧凑的提示证据进行检索。该设计支持在零样本设置下无需监督标签即可进行任务自适应的证据选择。在十个肿瘤学分类任务和一个MedQuAD CancerGov问答基准上，TRACE同时提升了无标签评估和监督微调的性能。进一步分析表明，TRACE优于普通RAG和通用GraphRAG，在泄漏控制的METABRIC输入下仍保持有效性，并产生与临床推理一致的可解释证据路径。这些结果表明，明确且可更新的医学结构是实现更准确、可审计的肿瘤学大语言模型部署的实用途径。
英文摘要
Large language models are increasingly used in oncology applications, but their predictions are often weakly grounded in explicit medical structure. We present TRACE, a deployable tree-relational enhancement framework for oncology LLMs. TRACE separates expensive offline structure learning from lightweight online inference: oncology concepts and relations are organized into an updatable tree-relational structure, refined using LM-loss-derived evidence, and retrieved at inference time as compact prompt evidence. This design supports task-adaptive evidence selection without requiring supervised labels in the zero-shot setting. Across ten oncology classification tasks and one MedQuAD CancerGov QA benchmark, TRACE improves both label-free evaluation and supervised fine-tuning. Additional analyses show that TRACE improves over vanilla RAG and generic GraphRAG, remains useful under leakage-controlled METABRIC inputs, and produces interpretable evidence paths aligned with clinical reasoning. These results suggest that explicit, updatable medical structure is a practical path toward more accurate and auditable oncology LLM deployment.
发表机构
- Pharmaron（康龙化成）
- AI Starfish
机构由 AI 辅助整理，请以论文原文为准。
