Generative models in traditional Chinese medicine: Advances and applications.
- 1Beijing Institute of Radiation Medicine, Beijing 100850, China.
- 2School of Pharmacy, Guangdong Pharmaceutical University, Guangzhou 510006, China.
- 3Xiyuan Hospital, China Academy of Chinese Medical Sciences, Beijing 100091, China.
Abstract
Generative artificial intelligence is increasingly being applied across interdisciplinary fields, enabling cross-modal content generation-spanning text, images, and molecular structures-through self-supervised learning. Confronting persistent challenges in Traditional Chinese Medicine (TCM), such as compositional complexity, ambiguous mechanisms, and unstructured knowledge systems, generative models may provide useful approaches by bridging fragmented data sources and offering scalable tools for prescription generation, molecular design, and knowledge discovery. In practice, large language models (LLMs) convert classical texts and electronic records into structured knowledge for syndrome-aware prescription generation, while fragment-aware GPT-based chemical language models expand natural-product chemical space. This review systematically explores core paradigms and recent advances in applying generative models to TCM research, focusing on key domains including TCM knowledge Q&A, prescription generation, and molecular design. We further critically analyze methodological and ethical challenges while outlining future directions in the field, such as interpretable generative models, integration with TCM knowledge graphs, and novel multimodal training strategies.
Related
- Traditional Chinese Medicine-Associated Bioactive Hydrogels: Material Roles, Network Design, and Therapeutic Applications.
- Mechanisms of traditional Chinese medicine interventions in the gastric ‘inflammation-to-carcinoma’ cascade: advances in in vitro models from monolayer cells to organoids
- Latent Profiles of Attitudes Toward Artificial Intelligence Applications Among Traditional Chinese Medicine Nurses and Associated Factors: A Multicenter Cross‐Sectional Study
- Applications of artificial intelligence in traditional Chinese medicine tongue diagnosis: a bibliometric review of research trends and hotspots, 2015-2026.
- Evidence quality assessment of traditional Chinese medicine and integrated traditional Chinese and Western medicine for acute upper respiratory infection in children: an umbrella review of systematic reviews and meta-analyses
BibTeX
@article{sun2026generative,
title = {{Generative models in traditional Chinese medicine: Advances and applications.}},
author = {Dezhi Sun and Cheng Zhang and Guangsen Li and Yangyi Hu and Shuman Li and Wei Zhou and Yue Gao},
journal = {Chinese herbal medicines},
year = {2026},
month = aug,
doi = {10.1016/j.chmed.2026.08.013},
url = {https://europepmc.org/article/MED/42948720},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Generative models in traditional Chinese medicine: Advances and applications.
AU - Dezhi Sun
AU - Cheng Zhang
AU - Guangsen Li
AU - Yangyi Hu
AU - Shuman Li
AU - Wei Zhou
AU - Yue Gao
PY - 2026
DA - 2026/08/25
JO - Chinese herbal medicines
DO - 10.1016/j.chmed.2026.08.013
UR - https://europepmc.org/article/MED/42948720
AB - Generative artificial intelligence is increasingly being applied across interdisciplinary fields, enabling cross-modal content generation-spanning text, images, and molecular structures-through self-supervised learning. Confronting persistent challenges in Traditional Chinese Medicine (TCM), such as compositional complexity, ambiguous mechanisms, and unstructured knowledge systems, generative models may provide useful approaches by bridging fragmented data sources and offering scalable tools for prescription generation, molecular design, and knowledge discovery. In practice, large language models (LLMs) convert classical texts and electronic records into structured knowledge for syndrome-aware prescription generation, while fragment-aware GPT-based chemical language models expand natural-product chemical space. This review systematically explores core paradigms and recent advances in applying generative models to TCM research, focusing on key domains including TCM knowledge Q&A, prescription generation, and molecular design. We further critically analyze methodological and ethical challenges while outlining future directions in the field, such as interpretable generative models, integration with TCM knowledge graphs, and novel multimodal training strategies.
SN - 1674-6384
SN - 2589-3610
ER -