Related Experiment Video
Updated: Apr 2, 2026
10:33
Exploring the Two Herb Combination Strategy to Treat Injured PC12 Cells
Published on: November 18, 2022
1.8K
TCMNet: an AI-driven strategy for optimizing traditional Chinese medicine
Shuoyan Tan1,2, Xin Shao3,4, Xuting Zhang2
1State Key Laboratory of Chinese Medicine Modernization, Pharmaceutical Informatics Institute, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou, 310058, China.
Chinese Medicine
|April 1, 2026
Summary
TCMNet, an AI strategy, enhances traditional Chinese medicine (TCM) formula evaluation by integrating large language models (LLMs) and protein-protein interaction (PPI) networks. This AI approach identifies key active compounds and optimizes TCM for modern medicine.
Area of Science:
- Computational biology and bioinformatics
- Artificial intelligence in medicine
- Pharmacology and drug discovery
Background:
- Traditional Chinese Medicine (TCM) complexity poses challenges for systematic analysis.
- Existing methods for TCM formula design and target prioritization lack objectivity.
- Artificial intelligence (AI), including Large Language models (LLMs), offers new tools for modeling TCM.
Purpose of the Study:
- To develop TCMNet, an AI-powered strategy for evaluating TCM herbal formulas and identifying active compounds.
- To overcome limitations of subjective design and unweighted target prioritization in TCM research.
- To integrate LLM-assisted knowledge mining, protein-protein interaction (PPI) networks, and deep learning for TCM analysis.
Main Methods:
- TCMNet integrates AI-guided literature analysis with weighted PPI network evaluation, using Parkinson's disease (PD) as a case study.
- A TCM-specific LLM (TCMChat) semantically weighted disease-associated protein targets.
- Herb-specific data generated weighted herb-related proteins, incorporated into a PPI network for node weighting.
- Deep learning (Boltz-2) predicted binding probabilities between herbal compounds and PD proteins.
Main Results:
- Weighted proximity metrics significantly outperformed unweighted measures in evaluating TCM formulas.
- Tianma Gouteng Decoction showed superior performance in target coverage and network proximity.
- Integrative strategies combining TCM with Levodopa enhanced network proximity compared to monotherapy.
- Flavonoids and isoflavonoids from Ginkgo biloba were identified as key anti-PD compounds, validated by deep learning predictions.
Conclusions:
- TCMNet provides an AI-driven strategy for optimizing TCM by incorporating protein weights via LLM-guided target identification and node-weighted evaluation.
- The approach facilitates herbal formula evaluation, optimization, and identification of bioactive constituents.
- TCMNet advances the modernization of herbal medicine research through systematic, data-driven analysis.
