Objective: The recent surge in large language models (LLMs) across various fields has yet to be fully realized in traditional Chinese medicine (TCM). This study aims to bridge this gap by developing a large language model tailored to TCM knowledge, enhancing its performance and accuracy in clinical reasoning tasks such as diagnosis, treatment, and prescription recommendations. Materials and Methods: This study harnessed a wide array of TCM data resources, including TCM ancient books, textbooks, and clinical data, to create 3 key datasets: the TCM Pre-trained Dataset, the Traditional Chinese Patent Medicine (TCPM) Question Answering Dataset, and the Spleen and Stomach Herbal Prescription Recommendation Dataset. These datasets underpinned the development of the Lingdan Pre-trained LLM and 2 specialized models: the Lingdan-TCPM-Chat Model, which uses a Chain-of-Thought process for symptom analysis and TCPM recommendation, and a Lingdan Prescription Recommendation model (Lingdan-PR) that proposes herbal prescriptions based on electronic medical records. Results: The Lingdan-TCPM-Chat and the Lingdan-PR Model, fine-tuned on the Lingdan Pre-trained LLM, demonstrated state-of-the art performances for the tasks of TCM clinical knowledge answering and herbal prescription recommendation. Notably, Lingdan-PR outperformed all state-of-the-art baseline models, achieving an improvement of 18.39% in the Top@20 F1-score compared with the best baseline. Conclusion: This study marks a pivotal step in merging advanced LLMs with TCM, showcasing the potential of artificial intelligence to help improve clinical decision-making of medical diagnostics and treatment strategies. The success of the Lingdan Pre-trained LLM and its derivative models, Lingdan-TCPM-Chat and Lingdan-PR, not only revolutionizes TCM practices but also opens new avenues for the application of artificial intelligence in other specialized medical fields. Our project is available at https://github.com/TCMAI-BJTU/LingdanLLM.

Benefiting from the development of network pharmacology, Traditional Chinese Medicine (TCM) shows great potential in modern drug discovery. Recently, more and more TCM-related databases have been established for both academic and industry research, but they are still insufficient in data standardization, integrity, and precision. To better accelerate the TCM research and overcome these shortcomings, we construct a web-based TCM platform, LTM-TCM, which is currently the most comprehensive TCM database that includes the following advantages: (1) High-quality data integration from fourteen TCM authoritative databases, especially with additional manual collected 41,025 clinical treatment records and 213 ancient Chinese medical books. (2) Accurate correction of multi-source TCM interactions (between symptoms, prescriptions, herbs, ingredients and targets) through in-house Biomedical Natural Language Processing (BioNLP) approaches in more than 30 million articles. (3) Diverse cross-field pipelines (e.g., bioactive ingredients screening, targets prediction, and mechanism prediction, etc.) help integrating traditional medicine with modern science in common aspects at both the molecular and phenotypic levels. In summary, LTM-TCM contains 1928 symptoms, 48,126 prescriptions, 9122 plants, 34,967 ingredients, 13,109 targets and 1170,133 interactions among all TCM related components. LTM-TCM has both Chinese and English interfaces, and it is accessible at http://cloud.tasly.com/#/tcm/home.

Background: Gliomas, the most prevalent type of primary brain tumor, stand out as one of the most aggressive and lethal types of human cancer. Methods & Results: To uncover potential prognostic markers, we employed the weighted correlation network analysis (WGCNA) on the Chinese Glioma Genome Atlas (CGGA) 693 dataset to reveal four modules significantly associated with glioma clinical traits, primarily involved in immune function, cell cycle regulation, and ribosome biogenesis. Using the least absolute shrinkage and selection operator (LASSO) regression algorithm, we identified 11 key genes and developed a prognostic risk score model, which exhibits precise prognostic prediction in the CGGA 325 dataset. More importantly, we also validated the model in 12 glioma patients with overall survival (OS) ranging from 4 to 132 months using mRNA sequencing and immunohistochemical analysis. The analysis of immune infiltration revealed that patients with high-risk scores exhibit a heightened immune infiltration, particularly immune suppression cells, along with increased expression of immune checkpoints. Furthermore, we explored potentially effective drugs targeting 11 key genes for gliomas using the library of integrated network-based cellular signatures (LINCS) L1000 database, identifying that in vitro, both torin-1 and clofarabine exhibit promising anti-glioma activity and inhibitory effect on the cell cycle, a significant pathway enriched in the identified glioma modules. Conclusions: In conclusion, our study provides valuable insights into molecular mechanisms and identifying potential therapeutic targets for gliomas.

The research on the direct target of traditional Chinese medicine (TCM) is the key to study the mechanism and material basis of TCM, but there is still no effective technical methods at present. For Compound Danshen dropping pills (CDDP), there is no report about its direct targets. In this study, the direct targets of CDDP were studied for the first time, especially focusing on the protein kinase family, which plays causal roles in a variety of human disease. Firstly, the literature database of CDDP was constructed by literature retrieval, and the important components contained in CDDP were extracted. Secondly, the potential direct targets of important components was obtained through querying public database and predicted by Multi-voting SEA algorithm. Then, the KinomeX system was used to predict and to filter the potential kinase targets of CDDP. Finally, the experimental verification was carried out. In total, 30 active kinase targets was obtained at 25 μg/ml concentration of CDDP, and 9 dose-dependent targets were obtained at 250 μg/ml concentration of CDDP. This is an efficient and accurate strategy by integrating the targets recorded in several public databases and the targets calculated by two in silico modelling approaches predict potential direct targets of TCM, which can lay an important foundation for the study of the mechanism and material basis of them, promoting the modernization of TCM.

Hypoxia is the main survival challenge that human beings encounter in high altitudes, it is also the leading cause of Acute Mountain Sickness (AMS). Studies have shown that hypoxia induces a large number of reactive oxygen species (ROS) in AMS patients, and the surge of ROS leads to the reduction of oxygen delivery capacity of erythrocyte, senescence and inflammatory impairment of erythrocyte and vascular endothelial cells to a certain extent. Through depicting a target-pathway network, our study indicates that Compound Danshen Dropping Pills (CDDP), which is one of the best-known traditional Chinese medicine used for the treatment of myocardial ischemic diseases, can improve red blood cell oxygen delivery capacity in AMS patients, alleviate tissue and organ damage, relieve a series of clinical symptoms caused by hypoxia through ROS clearance and related mechanisms. We further elucidate the active ingredients of CDDP targeting ROS related pathway by target-ingredient correspondence analysis. Tanshinone IIA, catechol and some other compounds of CDDP were identified to have certain targeting effect on ROS and ROS dependent pathways. This study provides new understandings of CDDP in clinical application on AMS.