摘要
针对肝细胞癌（HCC）现有分期系统忽略异质性的问题，本文提出 HCC-STAR，一种临床对齐的大语言模型。该模型读取电子病历叙事，联合输出基于风险分数的分期、排序一致的治疗方案及生存估计。研究构建了约 3 万例数据的增强语料库，并开发了知识对齐的推理框架，利用可验证的复合奖励进行优化。在多中心队列中，HCC-STAR 在治疗推荐和风险分层上优于临床指南及主流模型，显著提升了生存预测准确性，被专家评估为可信的决策支持系统。
AI 推荐理由
论文核心提出临床推理框架，优化可验证的推理步骤以超越指南记忆。
研究机构
Department of Computer Science & Technology, Tsinghua University, Beijing, China
Hepato-Pancreato-Biliary Center, Beijing Tsinghua Changgung Hospital, Key Laboratory of Digital Intelligence Hepatology (Ministry of Education), School of Clinical Medicine, Tsinghua Medicine, Tsinghua University, Beijing, China
School of Biomedical Engineering, Tsinghua University, Beijing, China
College of AI, Tsinghua University, Beijing, China
Hubei Provincial Key Laboratory of Portal Hypertension & Cirrhosis, Xiangta People's Hospital, Xiangta, China
Department of Hepatobiliary and Pancreatic Surgery, The Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China
Division of Hepatobiliary and Pancreas Surgery, Department of General Surgery, Shenzhen People's Hospital (The First Affiliated Hospital, Southern University of Science and Technology, The Second Clinical Medical College, Jinan University), Shenzhen, China
Department of Hepatobiliary Surgery, Southwest Hospital, Third Military Medical University (Army Medical University), Chongqing, China
Department of Hepatopancreatobiliary Surgery, Second Affiliated Hospital of Harbin Medical University, Harbin, China
Department of Hepatobiliary Surgery, The First Affiliated Hospital of Guangxi Medical University, Nanning, China
Department of Hepatopancreatobiliary Surgery, The Fourth Affiliated Hospital, Harbin Medical University, Harbin, China
Department of Liver Surgery, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College (CAMs & PUMC), Beijing, China
Department of Hepatobiliary Surgery, Tianjin Medical University Cancer Institute and Hospital, National Clinical Research Center for Cancer, Tianjin Key Laboratory of Digestive Cancer, Tianjin’s Clinical Research Center for Cancer, Tianjin, China
Department of Hepatobiliary & Pancreatic Surgery, Zhengzhou Hospital of Wuhan University, Wuhan, China
Clinical Medicine Research Center for Minimally Invasive Procedure, Wuhan University, Wuhan, China
Department of Hepatopancreatobiliary Surgery, Mengchao Hepatobiliary Hospital of Fujian Medical University, Fuzhou, China
Biological Information BioBank, Mengchao Hepatobiliary Hospital of Fujian Medical University, Fuzhou, China
Department of Radiotherapy, Beijing Tsinghua Changgung Hospital, School of Clinical Medicine, Tsinghua Medicine, Tsinghua University, Beijing, China
Pancreas Center, Tianjin Medical University Cancer Institute and Hospital, National Clinical Research Center for Cancer, State Key Laboratory of Drugability Evaluation and Systematic Translational Medicine, Tianjin Key Laboratory of Digestive Cancer, Tianjin’s Clinical Research Center for Cancer, Tianjin, China
These authors contributed equally to this work.
Corresponding authors: dongxipeng@tsinghua.edu.cn, hujihui@tjmu.com, dcsyf@tsinghua.edu.cn, dongjiahong@mail.tsinghua.edu.cn
论文信息