LLM · News hub
TCM AI
A living hub of Traditional Chinese Medicine large language model news, papers, surveys, open weights, and datasets. Synced with Awesome-TCM-LLM.
What is a TCM LLM?
A TCM LLM (Traditional Chinese Medicine large language model) is a large language model trained or fine-tuned for the domain of Traditional Chinese Medicine. These models are typically built on general-purpose Chinese foundation models (such as Qwen, ChatGLM, Baichuan, or LLaMA) and further adapted with domain corpora — classical TCM canons, renowned physicians' case records, herbal formula knowledge, clinical guidelines, and consultation dialogues — acquiring capabilities such as syndrome differentiation (辨证论治), herbal formula recommendation, TCM knowledge question answering, and understanding of the four diagnostic methods (inspection, listening/smelling, inquiry, and palpation).
Since the first wave of TCM LLMs in 2023 — BianQue, HuaTuoGPT, and ShenNong-TCM-LLM — the field has grown rapidly, now spanning intelligent consultation, clinical decision support, TCM education, licensing-exam evaluation, and knowledge-graph construction, including multimodal systems such as ShizhenGPT that handle tongue and facial imagery. This page tracks model releases, academic papers, benchmarks, and open datasets in the TCM LLM space.
Representative TCM LLMs
- BianQue (扁鹊): proactive-health large language model for Chinese living spaces (2023)
- HuaTuoGPT: large language model trained on Chinese medical corpora (2023)
- ShenNong-TCM-LLM (神农): large-scale TCM language model with instruction data and open weights (2023)
- ZhongJing (仲景): expert-knowledge-guided TCM LLM with vertical-domain fine-tuning (2025)
- Qibo (岐伯): continued pretraining + SFT for syndrome differentiation and QA, with the Qibo Benchmark (2025)
- ShizhenGPT: multimodal TCM LLM supporting the four diagnostic methods (2025)
- Baize-TCM-LLM (白泽): Qwen3-based TCM QA model series from the Institute of Chinese Materia Medica, CACMS (2025)
- TCMChat: generative TCM LLM with a 600k-sample herbal-knowledge dialogue dataset (2025)
- XuanHuGPT (悬壶): parameter-efficient fine-tuned TCM domain LLM (2025)
- MedChatZH: first-author work — a Baichuan-7B model fine-tuned for Chinese medical / TCM consultation
See the full catalog below — filterable by type and year.
TCM LLM benchmarks & datasets
- TCMBench: comprehensive benchmark for evaluating LLMs in the TCM domain
- MTCMB: multi-task TCM benchmark — 12 subsets, ~7,100 samples covering knowledge, reasoning, formulas, and safety
- TCM-Eval: evaluation dataset for TCM LLMs
- TCMChat-dataset-600k: 600k herbal-knowledge dialogue samples for fine-tuning
- ShenNong_TCM_Dataset: instruction data released with the ShenNong model
- Survey of key technologies for TCM LLMs (IJPRAI): systematic review of knowledge organization, assisted diagnosis, and clinical decision support
What you will find
- Selected news on products, policy, and open releases
- Open-weight and multimodal TCM LLM systems
- Benchmarks, evaluation suites, and surveys
- Pretraining / SFT datasets and related resources
Catalog
Updated: 2026-08-09
All resources
TCM LLM news, papers, surveys, datasets, and open weights — filter by type and year.
从大语言模型到智能体（兰州大学学报医学版综述）
Chinese-language systematic review organized around the LLM-to-agent transition for TCM clinical assisted diagnosis and treatment (J. Lanzhou Univ. Med. Sci. 2026;52(4):49-57).
Beyond the Poetic Bard（中医AI翻译评论）
Beyond the Poetic Bard: a perspective on accuracy, epistemology, and medical-context limits of generative-AI translation of TCM texts (Translation Review).
Hybrid Retrieval + Re-ranking TCM Prescription Generation
Hybrid retrieval with re-ranking to enhance LLM-based TCM prescription generation (Springer CCIS conference paper).
TCM-RobustSDT
TCM-RobustSDT: a robustness benchmark dataset for LLM clinical reasoning in TCM (Figshare).
Agentic and Knowledge-Grounded LLMs in TCM（预注册）
OSF preregistration (not a completed review) of a systematic review on agentic and knowledge-grounded LLMs in TCM: evidence mapping, text mining, and translation readiness.
Patient-Conditioned Dual Hypergraph Reasoning
Patient-conditioned dual-hypergraph reasoning for auditable TCM prescription support, organizing symptom/tongue/pulse evidence around patterns and treatment principles (Tianjin University).
Insightful Eye presents Bianshi Cloud TCM at WAIC 2026, built on a registered Bianshi multimodal model with four-diagnosis devices and assisted-care systems.
Andun Health debuts a seven-diagnosis TCM robot at WAIC 2026, integrating face/IR face/tongue/ear/auscultation/inquiry/pulse sensing with TianHui pulse algorithms and a TCM clinical LLM.
Guizhou Medical University and partners launch HuaZu BenCao, a national ethnic-medicine AI platform built on ShuZhi QiHuang + Qwen integrating multi-ethnic materia medica classics.
Shanghai Seventh People's Hospital (SHUTCM) releases the Qiyuan TCM LLM with pretraining, domain fine-tuning, and expert RL; supports generative medical records and master-physician Agent digital twins; showcased at WAIC.
中医大模型关键技术综述（IJPRAI）
Survey of key technologies for TCM LLMs: knowledge organization, aided diagnosis, and clinical decision support (formally published in IJPRAI, World Scientific).
DeepTCM1.0
DeepTCM1.0: a multi-expert AI agent built on general LLMs for interpreting the mechanisms of TCM compound formulas (Research Square preprint).
Evidence-Based TCM Visualization Diagnosis System
Evidence-based TCM visualization diagnosis system: Neo4j knowledge graph (241 patterns, 1,263 symptoms) with four-stage symptom matching (LLM-verified) and information-gain-driven active inquiry.
Zhongke Wenge passes HKEX listing hearing; reports note the DaYi JinKui TCM LLM (with CACMS) has obtained top-tier CAICT Trusted AI certification.
RAG+LoRA 中医执照考试推理架构
RAG+LoRA generative architecture with an 11,476-item Taiwan TCM licensing-exam dataset (2005-2025), raising accuracy from 61.0% to 89.0%+ (Future Internet, MDPI).
AI驱动中医诊断智能化综述（JTCMS）
Survey on multimodal fusion and LLMs for intelligent four-diagnosis in TCM: applications, challenges and outlook (JTCMS).
儿童流感中成药推荐系统（KG+LLM）
Knowledge graph of Chinese patent medicines for pediatric influenza built from authoritative guidelines and integrated with an LLM (JMIR Preprints).
HSQ-TD（健身气功指令微调数据集）
HSQ-TD: the first instruction-tuning dataset for health Qigong/wellness, with 57,843 instructions distilled from official textbooks and professional literature (ScienceDB).
病机推理CoT监督（脾胃病）
Pathogenesis-reasoning chain-of-thought supervision replacing fixed-label classification for spleen-stomach disease syndrome recognition and multi-dimensional evaluation (Prog. Biochem. Biophys.).
GAT+LLM TCM Prescription Generation
Intelligent TCM prescription generation combining graph attention networks with LLMs (formally published in KSII TIIS).
TCMIIES
TCMIIES: a browser-based, zero-installation LLM system for structured information extraction from academic literature, aimed at TCM and other specialty researchers.
Affiliated Hospital of Shandong University of TCM launches the provincial AI+ TCM scenario project ZhiHui QiHuang.
Beijing University of Chinese Medicine's XinHuo ZhongGuoYao TCM education LLM completes national generative-AI service filing—the first publicly approved TCM-domain model of its kind.
Eight ministries issue the TCM Industry High-Quality Development Plan (2026–2030), calling for AI and knowledge graphs to empower classical formulas and master physicians' prescriptions.
人工智能驱动下的中医智能诊疗研究进展与挑战
Chinese review structured on the six-step TCM diagnosis-treatment chain (four diagnoses, pattern differentiation, prescription, outcome prediction), contrasting supervised/unsupervised/RL/deep-learning paradigms (Shanghai J. TCM 2026;60(1)).
ZMT-M1 scores 96.26 on a simulated national TCM practitioner exam and is piloted in 100+ clinics.
Gushengtang launches a “TCM Brain” product and AI digital twin of National TCM Master Shi Qi; 14 expert twins cover eight core specialties.
Jilin releases ZhongXing·Changbai Qihuang 1.0, an AI-native multimodal TCM LLM.
Five national health ministries issue AI+ healthcare guidelines that explicitly support building TCM diagnostic LLMs.
YiYin classical TCM LLM goes live in Song County, Henan, combining TCM education with AI-assisted care for county clinics.
ZhiFu Qihuang Tiangong TCM AI model completes national deep-synthesis algorithm filing for four-diagnosis devices and constitution assessment.
Transn's RenDu·SuWen passes CAICT Trusted AI TCM LLM Level 4+ evaluation.
NSCC-TJ and Tianjin University of TCM release TianHe·LingShu 2.0 (expanded beyond acupuncture to 20+ specialties) and launch a TCM intelligent-model evaluation system.
Gushengtang releases ten “National Master AI twins” trained on master clinicians' experience, reporting >86% pattern-differentiation accuracy.
Guang'anmen Hospital forms the GuangYi·QiZhi LLM agent alliance with medical consortium partners for cross-institution intelligent care.
TCM Hengqin vertical LLM is officially released.
China Academy of Chinese Medical Sciences releases evaluation standards for TCM large models.
Transn releases RenDu·SuWen TCM LLM (mixture-of-entropy architecture) for inquiry, pattern differentiation, and formula recommendation.
Guang'anmen Hospital releases GuangYi·QiZhi, described as the first TCM hospital with integrated local compute + model + application deployment.
China UnionPay Consumer Finance with Sun Yat-sen University and GZUCMS Shenzhen Hospital release vertical TCM LLM ZhongSi for community clinic inquiry.
Zhongke Wenge releases DaYi JinKui TCM LLM and health platform trained on 1,500+ TCM classics.
Tasly and Huawei Cloud release ShuZhi BenCao (Pangu language + molecular models) covering TCM R&D; later earns CAICT TCM LLM Level 4+.
ECNU, SHUTCM, ECUST, NMMU, Lingang Lab, and CR Jiangzhong jointly develop the ShuZhi QiHuang TCM LLM.
Nanjing Dajing TCM releases QiHuang Wendao LLM with large knowledge graphs and clinical data for institutional beta use.
Cong H et al. TCM×LLM 综述（OSF 预印本）
OSF-preprint review of LLMs in TCM (not peer-reviewed; archival).
Cai R et al. TCM×LLM scoping review（OSF 预印本）
OSF-preprint scoping review of LLMs in TCM (not peer-reviewed; archival).
AI in TCM: multimodal data to pharmacology and clinical decision（PRMCM 综述）
Broad AI-in-TCM review from multimodal data integration to pharmacological research and clinical decision support (Pharmacol. Res. Mod. Chin. Med. 2026; found in the third-round scan).
TCMBank
Historical anchor: TCMBank.
ETCM v2.0
Historical anchor: ETCM v2.0.
ETCM
Historical anchor: ETCM.
TCMSP
Historical anchor: TCMSP.
TCMID
Historical anchor: TCMID.
Data-driven based four examinations in TCM: a survey
Historical anchor: Data-driven based four examinations in TCM: a survey.
Research and application of tongue and face diagnosis based on deep learning
Historical anchor: Research and application of tongue and face diagnosis based on deep learning.
Deep Learning Multi-label Tongue Image Analysis and Its Application in a Population Underg
Historical anchor: Deep Learning Multi-label Tongue Image Analysis and Its Application in a Population Underg.
Automatic Construction of Chinese Herbal Prescriptions From Tongue Images Using CNNs and A
Historical anchor: Automatic Construction of Chinese Herbal Prescriptions From Tongue Images Using CNNs and A.
Artificial intelligence in tongue diagnosis: Using deep convolutional neural network for r
Historical anchor: Artificial intelligence in tongue diagnosis: Using deep convolutional neural network for r.
Constitution Identification of Tongue Image Based on CNN
Historical anchor: Constitution Identification of Tongue Image Based on CNN.
Tooth-Marked Tongue Recognition Using Multiple Instance Learning and CNN Features
Historical anchor: Tooth-Marked Tongue Recognition Using Multiple Instance Learning and CNN Features.
Ensemble Learning-Based Pulse Signal Recognition: Classification Model Development Study
Historical anchor: Ensemble Learning-Based Pulse Signal Recognition: Classification Model Development Study.
Diagnostic Method of Diabetes Based on Support Vector Machine and Tongue Images
Historical anchor: Diagnostic Method of Diabetes Based on Support Vector Machine and Tongue Images.
Pulse Waveform Classification Using Support Vector Machine with Gaussian Time Warp Edit Di
Historical anchor: Pulse Waveform Classification Using Support Vector Machine with Gaussian Time Warp Edit Di.
Automated Tongue Feature Extraction for ZHENG Classification in Traditional Chinese Medici
Historical anchor: Automated Tongue Feature Extraction for ZHENG Classification in Traditional Chinese Medici.
Classification of Pulse Waveforms Using Edit Distance with Real Penalty
Historical anchor: Classification of Pulse Waveforms Using Edit Distance with Real Penalty.
Feature extraction and recognition of traditional Chinese medicine pulse based on hemodyna
Historical anchor: Feature extraction and recognition of traditional Chinese medicine pulse based on hemodyna.
A Novel Computerized Method Based on Support Vector Machine for Tongue Diagnosis
Historical anchor: A Novel Computerized Method Based on Support Vector Machine for Tongue Diagnosis.
An ontological framework for the formalization, organization and usage of TCM-Knowledge
Historical anchor: An ontological framework for the formalization, organization and usage of TCM-Knowledge.
Text mining for traditional Chinese medical knowledge discovery: a survey
Historical anchor: Text mining for traditional Chinese medical knowledge discovery: a survey.
Development of traditional Chinese medicine clinical data warehouse for medical knowledge
Historical anchor: Development of traditional Chinese medicine clinical data warehouse for medical knowledge .
Building Clinical Data Warehouse for Traditional Chinese Medicine Knowledge Discovery
Historical anchor: Building Clinical Data Warehouse for Traditional Chinese Medicine Knowledge Discovery.
Information retrieval and knowledge discovery on the semantic web of traditional Chinese m
Historical anchor: Information retrieval and knowledge discovery on the semantic web of traditional Chinese m.
Knowledge discovery in traditional Chinese medicine: State of the art and perspectives
Historical anchor: Knowledge discovery in traditional Chinese medicine: State of the art and perspectives.
Ontology development for unified traditional Chinese medical language system
Historical anchor: Ontology development for unified traditional Chinese medical language system.
Historical Analysis of Medical Artificial Intelligence Development in China: Research Cent
Historical anchor: Historical Analysis of Medical Artificial Intelligence Development in China: Research Cent.
Syndrome Differentiation in Intelligent TCM Diagnosis System
Historical anchor: Syndrome Differentiation in Intelligent TCM Diagnosis System.
Traditional Chinese medical diagnosis based on fuzzy and certainty reasoning
Historical anchor: Traditional Chinese medical diagnosis based on fuzzy and certainty reasoning.
Establishment of a fuzzy mathematical model for syndrome differentiation of gastric cancer
Historical anchor: Establishment of a fuzzy mathematical model for syndrome differentiation of gastric cancer.
Fuzzy match and floating threshold strategy for expert system in traditional Chinese medic
Historical anchor: Fuzzy match and floating threshold strategy for expert system in traditional Chinese medic.
关幼波肝病诊疗程序（肝病专家系统）
Historical anchor: 关幼波肝病诊疗程序（肝病专家系统）.
An artificial intelligence program to advise physicians regarding antimicrobial therapy
Historical anchor: An artificial intelligence program to advise physicians regarding antimicrobial therapy.
Mathematical modeling of Chinese medicine by complex-valued five-agent network
Historical anchor: Mathematical modeling of Chinese medicine by complex-valued five-agent network.
Discovering golden ratio in the world’s first five-agent network in ancient China
Historical anchor: Discovering golden ratio in the world’s first five-agent network in ancient China.
A disturbance rejection framework for the study of traditional Chinese medicine
Historical anchor: A disturbance rejection framework for the study of traditional Chinese medicine.
Equilibrium and nonequilibrium modeling of YinYang WuXing for diagnostic decision support
Historical anchor: Equilibrium and nonequilibrium modeling of YinYang WuXing for diagnostic decision support .
YinYang bipolar logic and bipolar fuzzy logic
Historical anchor: YinYang bipolar logic and bipolar fuzzy logic.
A computer model of the “five elements” theory of traditional Chinese medicine
Historical anchor: A computer model of the “five elements” theory of traditional Chinese medicine.
Functional structure model of human body and Yinyang-Wuxing equations
Historical anchor: Functional structure model of human body and Yinyang-Wuxing equations.
GPT vs ERNIE 中医文化背景对比研究
A culture-framed comparison of GPT versus ERNIE on TCM tasks (J. Integr. Complement. Med. 2024).
树状自反思检索中医问答
Tree-organized self-reflective retrieval for TCM question answering (Frontiers in Medicine 2026).
RACE-Align
RACE-Align: retrieval-augmented, CoT-style DPO alignment of a compact Qwen3-1.7B for TCM reasoning.
仲景（CMtMedQA 线，Yang et al.）
ZhongJing (CMtMedQA line, Yang et al.): a TCM LLM distinct from the Kang-line ZhongJingGPT—full CPT+SFT+RLHF pipeline on Ziya-LLaMA-13B over ~70K real multi-turn doctor-patient dialogues (AAAI 2024).
LLM-Based Multi-Agent Systems for Clinical Workflows（ACL 2026，邻近）
Adjacent ACL 2026 survey of workflow-level multi-agent clinical systems with a four-layer evaluation stack; no TCM coverage but methodologically isomorphic process-evaluation claims.
医学大语言模型的研发与应用系统综述（智能系统学报，邻近）
Adjacent systematic review of 129 medical-domain LLMs (to 2024-06) and four clinical application categories; methodologically comparable search protocol.
医疗领域的大型语言模型综述（智能系统学报，邻近）
Adjacent Chinese general survey of medical LLMs (training pipeline, strategies, scenarios, challenges), a superset-context reference for TCM LLM surveys.
Intelligent Question-Answering Systems in Healthcare（Healthcare，邻近）
Adjacent review (not TCM-specific): 2018-2025 healthcare QA survey with CiteSpace bibliometrics, explicitly covering TCM formula-development scenarios (Healthcare 2025).
人工智能实现中医四诊的发展现状、问题及解决路径（中华中医药学刊）
Short Chinese review of AI-based four-diagnosis objectification: face/tongue acquisition, electronic nose, pulse sensing, and low fusion of multi-diagnosis data (bibliographic record only).
人工智能赋能中医数字化诊断：现状与挑战（中华中医药学刊）
Short Chinese review of AI-empowered digital TCM diagnosis: applications, data-quality, interpretability, and theory-integration challenges (bibliographic record only).
AI for Spleen-Stomach Disorders in TCM（Curr Med Sci）
Single-disease-area (spleen-stomach) review of KG plus intelligent diagnosis/treatment with a symptom-syndrome-disease-formula framework (Curr. Med. Sci. 2025;45(6)).
AI and Big Data in TCM Standardization and Internationalization（Chin Med Cult）
Perspective on AI and big data for TCM standardization and internationalization (Chin. Med. Cult. 2026, ahead of print).
AI empowers the innovation of TCM（J Integr Med 评论）
Single-author perspective on AI for TCM innovation: classics mining, diagnosis standardization, drug R&D cycles (J. Integr. Med. 2026).
古籍知识图谱×多智能体融合综述（Chin Med）
Challenges-and-prospects review of knowledge-graph construction over ancient TCM classics, first to frame multi-agent convergence in this area (Chin. Med. 2025;20:168).
The integration of machine learning into TCM（J Pharm Anal）
Review of machine-learning integration into TCM along diagnostic objectification and mechanism-elucidation lines (J. Pharm. Anal. 2025;15(8):101157).
Deep learning in TCM（J Integr Med）
Single-technology review of deep learning in TCM: medical imaging, herbal material research, data mining (J. Integr. Med. 2026;24(4):471-480).
AI in TCM: Unraveling Herbal Medicine's Mechanisms（Research）
Broad AI-in-TCM review arguing AI should move beyond correlational analysis toward reconstructing the biological logic of syndrome differentiation and formula compatibility (Research 2026;9:1224).
多模态大模型驱动舌脉面诊智能化综述（Springer 书章）
The only review text dedicated to multimodal-LLM-driven tongue, pulse, and facial diagnosis in TCM (Springer CCIS book chapter; weaker peer review than journals).
知方丹台
ZhiFangDanTai formula-generation model weights.
白泽 (Baize)
Baize TCM LLM weights.
medchatzh
MedChatZH weights.
ZhongJing
ZhongJing GPT weights.
TCMChat
TCMChat weights.
ShizhenGPT
ShizhenGPT multimodal weight series.
ShenNong-TCM-LLM
ShenNong-TCM-LLM weights.
Lingdan
Lingdan / TCMLLM weights.
ChatTCM-7B-SFT
ChatTCM full-parameter SFT checkpoint.
ChatTCM
ChatTCM pretrained weights.
BianCang
BianCang open-weight series.
杏核 (Xinghe)
Xinghe Neijing reasoning model weights.
TCM_KG
ChatMed knowledge graph.
TCM-MKG
TCM multi-dimensional knowledge graph.
OpenTCM-KG
OpenTCM gynecology classics KG (~48k entities / ~152k relations).
TCMNSCLC
Real-world NSCLC TCM reasoning dataset with fully annotated cases (pattern differentiation / treatment method / decoction / patent medicine).
ChP-TCM
KnowledgeQA and PrescriptionWriting instructions built from Chinese Pharmacopoeia Vol. I.
Traditional-Chinese-Medicine-Dataset-SFT
High-quality TCM supervised fine-tuning dataset.
TCMChat-dataset-600k
TCMChat herbal QA and recommendation instruction data (~600k).
TCM-Instruction-Tuning-ShizhenGPT
ShizhenGPT multimodal SFT data (text/vision/speech/ECG etc.; ~311k items total per paper Table 3).
ShenNong_TCM_Dataset
ShenNong TCM instruction dataset.
MedChatZH
MedChatZH TCM consultation dataset.
ChatMed_Consult_Dataset
Chinese online medical consult dataset (500k+ consults with ChatGPT replies).
CMtMedQA
ZhongJing real multi-turn doctor–patient dialogues (~70k).
Baize-TCM-Corpus-V3
~157k TCM QA items covering theory, herbs, formulas, diagnosis, acupuncture, and clinic.
neijing-sft-v1.2
~2,009 Neijing-related instruction samples for Xinghe, with thinking/output fields.
TCM-Text-Exams
Recent TCM licensure / graduate-exam text benchmark.
Medical-LLMs-Chinese-Exam
Chinese medical exam evaluation for medical LLMs.
ZhongJing-OMNI
ZhongJing-OMNI multimodal TCM eval (including tongue).
TCMEval-SDT
TCMEval-SDT: a benchmark of 300 syndrome-diagnosis cases (web, classical texts, hospital records) for evaluating TCM syndrome-differentiation reasoning, with FAIR metadata (Sci. Data 2025).
TCMBench
TCMBench: a comprehensive benchmark for evaluating LLMs in traditional Chinese medicine (arXiv 2024).
TCM-Vision-Benchmark
TCM vision benchmark (herb recognition / inspection, ~7k items).
TCM-Tongue
6,719 standardized tongue images with 20-class multi-label pathology annotations and detection baselines.
TCM-Ladder
TCM-Ladder: a multimodal QA benchmark for comprehensively evaluating TCM multimodal LLMs on real-world tasks (arXiv 2025).
TCM-Eval
Dynamic, extensible TCM evaluation platform.
TCM-BEST4SDT
Case benchmark for syndrome differentiation and treatment.
TCM-5CEval
Five-dimension deep TCM evaluation suite.
TCM-3CEval
Three-axis eval: core knowledge, classics, clinical decisions.
MTCMB
MTCMB dataset: a multi-task TCM benchmark covering knowledge, reasoning and safety, 12 subsets with ~7,100 samples (arXiv 2025).
HWTCMBench
HWTCMBench TCM capability evaluation set.
TCMEval-PA
328 multiple-choice items on prescription normative quality and safety auditing.
LingLan
LingLan large multi-task TCM evaluation benchmark (2026).
ChiMed 2.0
Upgraded Chinese medical pretraining dataset covering TCM corpora for LLM pretraining.
classical-tcm-canon
Full-text digitizations of the TCM canon: Neijing, Nanjing, Shanghan Lun, Jingui Yaolue and warm-disease classics.
Traditional-Chinese-Medicine-Dataset-Pretrain
High-quality TCM pretraining dataset from non-Internet sources (~1GB; clinical cases, classics, encyclopedia), 99% simplified Chinese.
TCM-Pretrain-Data-ShizhenGPT
ShizhenGPT pretraining corpus (15B+ tokens reported in the paper — Stage-1 text 11.92B incl. 6.3B TCM, plus Stage-2 multimodal ~3.6B).
TCM-Ancient-Books
A corpus of nearly 700 TCM ancient-book texts.
awesome_Chinese_medical_NLP
Curated list of Chinese medical NLP resources: terminologies, corpora, word vectors, pretrained models, KGs, NER and QA (incl. CBLUE).
CPM中成药数据集
Living large-scale public Chinese patent medicine data accompanying RAG-CPMF.
Lukman et al. 2007: 中医计算方法综述
Foundational survey of computational methods for TCM (expert systems, ML, data mining).
Gu & Chen 2013: 生物信息学遇见中医
Historical review of bioinformatics meeting TCM (omics and text mining).
Zhao et al. 2015: 中医患者分类进展（ML 视角）
Review of ML-driven advances in patient classification for TCM.
Chu et al. 2020: 中医定量知识表示模型综述
Review of quantitative knowledge representation models of TCM (ontologies, rules, statistics).
Zhang et al. 2021: 计算中医诊断文献综述
Literature survey of computational TCM diagnosis — symptom acquisition, pattern modeling, and systems.
Tian et al. 2024: 四诊机器学习综述
Review of machine learning for TCM four diagnoses — inspection, auscultation-olfaction, inquiry, and palpation.
Qu et al. 2024: 中医知识图谱综述
Review of knowledge graphs in TCM — analysis, construction, applications, and prospects.
Song et al. 2024: AI 辅助中医辨证关键问题与技术挑战
Strategic-study review of key issues in AI-assisted TCM syndrome differentiation — multimodal fusion, symptom association, pattern quantification and reasoning, and TCM LLMs.
Su et al. 2024 — Review of AI in TCM diagnosis and treatment (Chinese)
Chinese-language review of three AI stages in TCM care — expert systems, ML, and deep learning — with challenges.
Li et al. 2024 — Research progress and prospects of LLMs in TCM (Chinese)
Chinese-language review of TCM LLM pipelines, frontier techniques (prompting/RAG/RLHF), and application prospects.
Yip et al. 2025: 中西医结合 LLM 进展与挑战
Review of LLMs in integrative medicine — progress, challenges, and opportunities.
Wang et al. 2025: AI 驱动中医诊断模型进展
Systematic review of AI-driven TCM diagnostic models (four-diagnosis objectification, pattern differentiation).
Meng et al. 2025: 大模型+虚拟细胞助力中医变革
Review of large models and virtual cells aiding modern analysis of stroke treatment with TCM formulas.
Zhang et al. 2025: 中医 LLM 短综述与展望
Short survey and outlook on TCM LLM models and tasks.
Shataer et al. 2025: LLM 在中医应用（State-of-the-Art Review）
State-of-the-art review scanning TCM LLM application scenarios (care, education, translation, research).
Guo et al. 2025: GPT 能否加速中医智能诊疗（综述+实证）
Survey plus empirical analysis of whether GPTs can accelerate intelligent TCM diagnosis and treatment.
Chen et al. 2025: 中医大语言模型系统综述
Systematic review of 10 studies (to mid-2024) on LLMs in TCM generative tasks.
Ren et al. 2025: 中医大语言模型（Scoping Review）
Arksey-O'Malley scoping review (29 studies to 2024-04) covering knowledge management, assisted care, and exam accuracy.
Lu et al. 2026: 深度学习中医诊断方法学质量审计
Systematic review and validation-gap analysis of deep learning for TCM disease diagnosis.
Wu et al. 2026: AI 在中药材中的应用综述
Full-stack survey of AI in TCM herbs — compounds, targets, quality control, with an LLM section.
Guo et al. 2026: AI 与多模态数据融合推动中医现代化
Panoramic AI review (ML/DL/KG/NLP/LLM) for TCM modernization with multimodal data integration.
Chen et al. 2026: LLM 在中医的下一步（叙述性综述）
Narrative review on the next step of LLMs in TCM — multimodality, agents, and clinical translation.
Xu et al. 2026: 基于 LLM 的中医智能问答系统综述
Review of intelligent TCM question-answering systems based on LLMs (KG-QA to LLM-QA and RAG).
Yao et al. 2026: LLM 与循证中医整合（Scoping Review）
PRISMA scoping review (12 studies, 2022-11 to 2026-01) on integrating LLMs with evidence-based Chinese medicine.
Han et al. 2026: LLM 在中医中的调优与临床应用（Scoping Review）
PRISMA-ScR scoping review (27 studies to 2025-05) on tuning (LoRA/CPT/RAG) and clinical application of TCM LLMs.
中医症状名识别
Supervised methods for symptom name recognition in free-text TCM clinical records.
中医临床细粒度 NER 语料
Fine-grained entity-recognition corpus built from TCM clinical records.
TCMKG
Deep-learning-based TCM knowledge graph platform.
乙肝中医 KG 问答系统
Knowledge-graph-based QA system for TCM diagnosis and treatment of viral hepatitis B.
中医新冠文献 LLM 命名实体识别
Comparative study of LLMs for named entity recognition in TCM COVID-19 literature (preprint).
PreGenerator
TCM prescription recommendation model combining retrieval and generation.
草药智能配送聊天机器人
Smarter herbal medication delivery system employing an AI-powered chatbot.
LLM+GNN 中医处方推荐
TCM prescription recommendation combining large language models with graph neural networks.
中医方剂 LLM 分类
Fine-tuned LLMs with refined prompt templates for TCM formula classification, using data sources such as the national medical-insurance catalog of proprietary Chinese medicines (IEEE BIBM 2023).
中医疫病防治问答模型
LLM-based QA model for TCM epidemic prevention and treatment.
ChatGPT 针灸教育研究
Comparative study of ChatGPT as a learning tool in acupuncture education.
大模型融合知识图谱问答系统
Vertical-domain QA system deeply integrating LLMs with knowledge graphs for TCM formulas.
黄帝 (HuangDi)
HuangDi: a TCM classics QA LLM built on Ziya-LLaMA-13B, pretrained on 22 TCM textbooks plus TCM web corpora and SFT-tuned with ancient-book instruction data (Library Tribune 2024).
神农大模型 (ShenNong-TCM-LLM)
ShenNong-TCM-LLM, the first TCM large language model, released with the ShenNong_TCM_Dataset and open weights.
BianQue
Chinese proactive health LLM for everyday living spaces (BianQue).
孙思邈 (Sunsimiao)
Sunsimiao Chinese medical LLM; Sunsimiao-7B fine-tuned from Qwen2-7B on curated medical data, reaching 30B-level SOTA on CMB-Exam.
QiZhenGPT
Chinese clinical QA model for drugs, diseases, procedures, and labs (QiZhenGPT).
HuaTuoGPT
Large language model trained on Chinese medical corpora (HuaTuoGPT).
XrayGLM
Chinese multimodal medical LLM for chest X-ray interpretation.
ChatMed
ChatMed series of Chinese medical LLMs, including ChatMed-Consult trained on 500k+ online consultation dialogues.
RAG 增强中医问答置信度
Implementing retrieval-augmented generation to build LLM confidence in TCM (preprint).
GPT-4 中医研究生考试评估
GPT-4 vs mainstream Chinese LLMs on a TCM postgraduate examination dataset (preprint).
中医药问答大语言模型
TCM QA LLM combining RAG with P-Tuning v2 fine-tuning on ChatGLM2-6B.
中医标准化评估基准
Standardized TCM evaluation benchmark of 29,506 questions across 13 subjects; tests 3 general and 5 Chinese medical LLMs.
中医药大模型知识增强方法
Knowledge augmentation for TCM LLMs — a graph built from ~100k classical formulas preserving prescription structure.
ACUBERT
ACUBERT for meridian entity recognition and classification in acupuncture indication knowledge bases.
BSG 中医智能问答
Intelligent QA system for TCM based on a BSG deep-learning model (prescription and materia medica cases).
TCMD
TCMD, a TCM QA dataset for evaluating large language models.
RLAIF 中医对齐
Enhancing LLMs' TCM capabilities through reinforcement learning from AI feedback.
中医提示工程框架
Prompt-engineering framework for LLM intelligent understanding in TCM.
ChatGPT 中医知识理解探究
Evaluating ChatGPT's comprehension of Traditional Chinese Medicine knowledge.
TCMSF
TCMSF: a construction framework for a TCM syndrome ancient-book knowledge graph that organizes syndrome knowledge from classical texts in a structured, semantically oriented way (Methods Inf. Med. 2024).
TCM MLKG-RAG
TCM intelligent diagnosis based on multi-layer knowledge graph retrieval-augmented generation.
Evi-BERT
Automated information-extraction model (Evi-BERT) enhancing RCT evidence extraction for TCM.
ChatGPT 中医交互可行性研究
Feasibility and challenges of interactive AI for TCM, using ChatGPT as an example.
中医领域知识图谱补全
Domain knowledge graph completion and quality evaluation for Traditional Chinese Medicine.
LLM 构建中医知识图谱
Constructing Traditional Chinese Medicine knowledge graphs based on large language models.
LLM 中医语言文化偏差研究
Comparing LLMs developed in different countries on TCM; highlights language/cultural bias and the need for localized models.
LLM 腧穴定位关系抽取
Relation extraction with LLMs — a case study on acupuncture point locations.
TCM-FTP
Fine-tuning LLMs for herbal prescription prediction.
CPMI-ChatGLM
Parameter-efficient fine-tuning of ChatGLM with Chinese patent medicine instructions.
TCM-GPT
Efficient pre-training of LLMs for domain adaptation in Traditional Chinese Medicine.
BenCao (formerly HuaTuo)
Instruction-tuned Chinese medical LLM (BenCao / formerly HuaTuo).
明医 (MING)
MING: a Chinese medical consultation LLM using a sparse mixture of low-rank adapter experts (MING-MoE) for medical multi-task learning (arXiv 2024).
大数中医 (BigDataTCM)
BigDataTCM (34B): a vertical TCM LLM co-developed by HAUT's Complexity Science institute and Apus, offering medical QA, diagnostic support and TCM knowledge services.
TCMLLM / Lingdan
TCMLLM / Lingdan for TCM modeling and prescription recommendation.
MedChatZH
MedChatZH: a fine-tuned LLM for TCM consultation dialogues, released with open dataset and weights (Comput. Biol. Med. 2024).
Chinese-LLaVA-Med
Chinese medical multimodal LLM based on the LLaVA architecture, with the llava-med-zh-eval benchmark and open 7B weights.
GPT 台湾中医执业考试评估
GPT-3.5/GPT-4/GPT-4o performance on the Taiwan TCM licensing examination with reliability analysis (preprint).
双通道知识注意力辨证模型
Dual-channel knowledge-attention NLP model for TCM syndrome differentiation, addressing rare characters and terminology extraction.
中医药标准知识问答系统
Retrieval-augmented QA system for TCM standards knowledge, built and evaluated in practice.
Hengqin-RA-v1
LLM and companion dataset for TCM diagnosis and treatment of rheumatoid arthritis.
Gen-SynDi
Knowledge-guided generative-AI framework for dual education of syndrome differentiation and disease diagnosis.
辨证思维评测 (Syndrome Differentiation Thinking)
Method-development study evaluating and improving LLMs' TCM syndrome-differentiation thinking ability.
针灸大模型驯化与生成评估 (Taming LLMs for Acupuncture)
Taming LLMs for acupuncture & moxibustion diagnosis, with generation quality evaluated at the semantic-similarity level.
TCM-Sage
Evidence-synthesis RAG assistant for TCM practitioners (hybrid vector + knowledge graph).
From Metaphor to Mechanism
LLMs decode TCM metaphor / imagistic-thinking language and map it to modern medical concepts.
New Snow Tablets
Reveals systematic flaws of general and TCM-specific LLMs that guess formula ingredients from drug names.
TCDiff
Triplet cascaded diffusion model generating high-fidelity multimodal TCM EHRs, with the TCM-SZ1 benchmark dataset.
Ladder-base (GRPO-TCM)
First GRPO reinforcement-learning-aligned TCM LLM, from the TCM-Ladder team.
MRD-RAG
Multi-round diagnostic RAG simulating clinical reasoning; builds DiagnosGraph spanning TCM and Western medicine (876 diseases / 7,997 nodes / 37,201 triples).
TCM compound retrieval agent
AI agent-based system for retrieving TCM compound information.
LM extraction for complementary medicine
Language models for data extraction and risk-of-bias assessment in complementary medicine literature.
LLM-driven TCM KG construction
LLM-driven construction and application of a TCM knowledge graph.
中医医案问答系统
A TCM case-based QA system integrating LLMs and knowledge graphs for efficient case retrieval and analysis (Front. Med. 2025).
LLM + RAG TCM inference
Combining LLMs with RAG for TCM inference.
Weighted-voting TCM formula classification
Weighted-voting LLM approach for TCM formula classification.
TCM guideline adherence evaluation
Content-analysis evaluation of LLM adherence to clinical practice guidelines in Chinese medicine.
5-LLM TCM clinical decision comparison
Comparative study of 5 LLMs for TCM clinical decision-making.
TCM stroke LLM benchmark
Quantitative benchmark study of LLMs in the TCM stroke domain.
LLM herb–drug interaction prediction
LLM-enhanced herbal medicine–drug interaction prediction.
TCMLCM
KG2T-based intelligent QA model for TCM lung cancer.
Chinese patent medicine knowledge system
Constructing a knowledge system for traditional Chinese patent medicine using LLMs and KGs.
TCMRD-KG
Innovative design of a rheumatology TCM knowledge graph from ancient literature.
TCM-Eval (WISE 2025)
Multi-dimensional TCM evaluation framework (WISE 2025); a different work from the ZMT-M1 TCM-Eval (arXiv 2511.07148) despite the identical name.
Few-shot tongue-diagnosis in-context multitask learning
Few-shot in-context multitask fine-tuning of LLMs mapping tongue images directly to constitutions.
TCM misinformation detection evaluation
Safety evaluation framework with 3,000+ TCM exam items × 4 paradigms, covering wrong-option, misleading, and fabrication detection.
ChatGLM-FGIDs-TCM
Knowledge-fused ChatGLM clinical decision-support model for functional gastrointestinal disorders (FGIDs).
TCM-VisResolve (TCM-VR)
Qwen2.5-VL-based TCM multimodal LLM — 163-class dried-herb recognition over 220k images plus clinical MCQs with 880k candidate answers.
TCM-DS
Domain LLM for medicine–food homology dietary-therapy recommendation.
XuanHuGPT
TCM domain LLM built with parameter-efficient fine-tuning (PEFT).
Jingfang
LLM-based multi-agent TCM diagnosis/treatment system reporting large relative SDT gains under the authors' protocol.
神农Alpha
ShennongAlpha (Westlake University): an AI-driven sharing and collaboration platform for intelligent curation, acquisition and translation of natural-medicinal-material knowledge (Cell Discov. 2025).
ZhiFangDanTai
GraphRAG + LLM fine-tuning for interpretable formula generation (sovereign–minister–assistant–courier, efficacy, contraindications) with open weights.
Baize-TCM-LLM
ICMM Baize TCM QA models on Qwen3 (0.6B/8B) with ~157k LoRA-tuning examples.
ZMT-M1
ZMT-M1 TCM LLM and the dynamic, extensible TCM-Eval benchmark platform.
BianCang
BianCang TCM LLM series (IEEE JBHI); 14B open-weight release in Dec 2025.
Qibo
TCM LLM and Qibo Benchmark from Tianjin University et al.; CPT + SFT for SDT and QA.
TianHui
Domain LLM for 12 TCM scenarios (DeepSeek-R1-Distill-Qwen-14B + PT/SFT) with open code and eval scripts.
Tianyi
~7B TCM LLM from NJUCM et al. with reading–clinic–apprenticeship training stages, TCMEval, and real-world validation.
仲景 (ZhongJing)
ZhongJingGPT, an expert-knowledge-guided TCM LLM combining vertical-domain fine-tuning with cognitive-psychology insights and multi-scenario TCM knowledge instructions (Tsinghua Sci. Technol. 2025).
RenShu-AI
FastAPI + LangGraph multi-agent TCM consultation system combining GraphRAG and DeepSeek-TCM.
Yaoshi-RAG
Uncertain-KG RAG for medicine–food homology dietary recommendation with personalization and explainability.
ViTCM-LLM
Qwen2.5-VL + RAG tongue multimodal clinical framework; MedTCM dataset and TDEU metric (precursor to MMIR-TCM).
TCMChat
Generative TCM LLM built via pre-training and supervised fine-tuning, released with the 600k-sample TCMChat-600k dialogue dataset (Pharmacol. Res. 2024).
TCM-R1
TCM LLM with GRPO-enhanced reasoning.
TCM-Ladder
First large multimodal TCM QA benchmark with 52,000+ items (NeurIPS 2025).
TCM-KLLaMA
KG-fused LLM for intelligent TCM formula generation.
TCM-BEST4SDT
Case benchmark for syndrome differentiation and treatment (knowledge / ethics / safety / SDT).
TCM-5CEval
Five-dimension deep evaluation extending TCM-3CEval with materia medica and non-drug therapies.
TCM-3CEval
Three-axis TCM LLM evaluation: core knowledge, classics comprehension, and clinical decision-making.
TCM LLM acupuncture clinical evaluation
Real-case evaluation of 7 general LLMs vs licensed acupuncturists on SDT, point selection, needling, and herbs (*npj Digital Medicine*).
ShizhenGPT
Multimodal TCM LLM supporting the four diagnoses (inspection, auscultation-olfaction, inquiry, palpation).
RAG-CPMF
Multi-LLM verification + RAG for Chinese patent medicine recommendation, with a living public CPM dataset.
OpenTCM
GraphRAG TCM retrieval and diagnosis system with a gynecology classics knowledge graph.
MTCMB
Multi-task TCM benchmark (~12 subsets, ~7.1k samples) covering knowledge, reasoning, formulas, and safety.
MCM
Multi-agent collaborative multimodal TCM diagnosis framework (IEEE ICIP 2025).
DoPI
Doctor-like proactive inquiry TCM LLM (guide + expert models); reported inquiry accuracy 84.68%.
DiagX-DT
Exclusionary syndrome-differentiation reasoning with CoT and an external TCM knowledge base.
ChatTCM
Fully open TCM LLM from pretraining data through released weights.
BenCao
Instruction-aligned multimodal TCM assistant (ChatGPT/GPTs Store) with tongue APIs and knowledge bases (distinct from HuaTuo/BenCao).
Medicinal-plant MLLM benchmark
Benchmarking multimodal LLMs for medicinal plant identification.
Large vs lightweight LLMs on TCM exams
Systematic comparison of large-scale vs lightweight LLMs on TCM exam questions.
TCM AI-tutor evaluation
Multimodal LLM evaluation for TCM education across cognitive levels.
Three-LLM TCM licensing exam evaluation
Systematic evaluation of 3 LLMs (incl. Gemini) on the national TCM medical licensing examination.
TCM intelligent pre-consultation clinical evaluation
Tertiary-hospital clinical evaluation of an LLM intelligent pre-consultation system using a physician–AI–patient triad model.
TCM Data Hub (YiYuan)
YiYuan LLM-driven TCM data platform.
TCMNet
LLM-assisted disease knowledge mining with PPI networks and binding prediction for formula optimization.
CMM-EmbedCluster
LLM + medicinal-property-theory clustering framework for Chinese materia medica, with a 567-herb property knowledge base.
KDC-NER
Knowledge-guided data augmentation + LLM fine-tuning framework for nested NER in TCM.
Jin San Zhen KG-QA
Knowledge graph + LLM QA tool for the Jin San Zhen acupuncture school.
TCMI-F-6D
Six-dimensional benchmark of interdisciplinary foundational competence in TCM informatics.
Tongue–face multimodal fusion diagnosis
Tongue–face multimodal feature fusion with LLM-driven intelligent TCM diagnosis.
TCMBenchEval
Benchmark evaluating LLMs on real clinical TCM case records (ICIC 2026).
Med-Bench-Arena
Open evaluation platform for medical and TCM LLMs/Agents (HF/vLLM/LiteLLM, multimodal, TCM-specific metrics), from the ZhongJing team.
TongueDx2
Systematic ablation of the tongue-diagnosis DL design space (20+ model variants); TongueDx2 includes 5,109 images / 976 expert annotations.
MACAT
Multi-agent culture-aware translation framework, evaluated on culture-loaded terms from TCM classics and the Analects.
DongYuan
Integrative spleen–stomach disease diagnosis LLM framework combining TCM pattern differentiation with Western diagnostic reasoning.
Med-Shicheng
Lightweight master-physician experience-inheritance framework built on Tianyi; a single model internalizes 5 national masters' knowledge systems across 7 task types.
HerbWise
Domain LLM for traditional herbal medicine (THM), serving herbal modernization and standardization.
QingNangTCM
Parameter-efficient fine-tuned TCM QA and clinical reasoning model; builds the 100k-item QnTCM_Dataset.
End-to-end TCM clinical support benchmark
Benchmark for end-to-end TCM clinical support across the full LLM care pipeline.
ATCMD-Bench
First agentic TCM diagnosis benchmark, evaluating LLMs through multi-agent simulated consultations.
LingLan
Large multi-task TCM benchmark: 5 domains, 13 subtasks, 25,624 instances.
Xinghe
Qwen3.5-9B reasoning TCM model grounded in the *Neijing*, with explicit CoT pattern differentiation and safety boundaries.
TongueVLM
Multimodal VLM for TCM tongue diagnosis, description generation, and constitution reasoning.
TCM-DiffRAG
Syndrome-differentiation RAG with a general KG, a personalized KG, and chain-of-thought.
TCM-Agent
LLM multi-agent system for network pharmacology and herbal discovery.
Qwen-TCM-Dia
Specialty fine-tuned model for TCM diarrhea care (CPT + CoT SFT) covering symptom→pathomechanism→method→formula chains.
MMIR-TCM
Memory-augmented multimodal tongue diagnosis and clinical decision framework; proposes MedTCM dataset and TDEU metric.
GastroTCM
TCM gastroenterology LLM fine-tuned from Llama3-8B with RAG and agent scaffolding.
DERM-3R
Resource-constrained multimodal multi-agent framework for TCM dermatology (recognition / representation / SDT agents).
CORE-Acu
Acupuncture clinical decision support with structured reasoning traces and a knowledge-graph safety veto loop.
No matching resources.
FAQ
What is a TCM LLM?
A TCM LLM is a large language model adapted to Traditional Chinese Medicine through continued pretraining or instruction tuning on TCM canons, case records, herbal formulas, and clinical corpora. It can perform syndrome differentiation, formula recommendation, and TCM knowledge QA. Representative systems include BianQue, HuaTuoGPT, ShenNong, ZhongJing, and ShizhenGPT.
What are the representative TCM LLMs?
Representative TCM LLMs include BianQue, HuaTuoGPT, ShenNong-TCM-LLM, ZhongJing, Qibo, ShizhenGPT, Baize-TCM-LLM, TCMChat, XuanHuGPT, and MedChatZH (first-author work on this site). See the catalog above for the full list.
What public TCM LLM datasets and benchmarks are available?
Widely used benchmarks include TCMBench, MTCMB, TCM-Eval, and the Qibo Benchmark; open datasets include TCMChat-dataset-600k, ShenNong_TCM_Dataset, and the TCM-Ancient-Books corpus. Filter by "dataset" or "benchmark" in the catalog.
How often is this TCM LLM list updated?
The catalog is maintained in sync with the open-source GitHub project Awesome-TCM-LLM; newly released TCM models, papers, benchmarks, and datasets are added continuously. The last-updated date is shown at the top of the catalog.
How can I submit a new TCM LLM resource?
Suggest models, papers, datasets, or news via a GitHub Issue; accepted submissions are synced to this page and the README.
Related first-author work
MedChatZH is a Baichuan-7B model fine-tuned for Chinese medical / TCM consultation — a concrete model release alongside this curated hub.
Contribute
Suggest a resource via GitHub Issue. Star the project on GitHub if you find it useful. Chinese UI: /zh/projects/tcm/.