Use Text Mining and Complex Networks to Analyze TCM Syndromes

Xing Zhai, Ping Wang, Youliang Huang2017 · DOI

Abstract

Objective: To make intrinsic biological distinctions between qi deficiency pattern and qi stagnation pattern with genes related to NEI. Methods: Establish a data dictionary of NEI-related genes and a keyword lexicon of qi deficiency pattern and qi stagnation pattern. Then retrieve relevant literature on Pubmed database. Obtain compositions of characteristic NEI-related genes of qi deficiency pattern and qi stagnation pattern by using text-mining method to explore different bioactive materials between the two syndromes. Results: Two kinds of syndromes and their biological networks were constructed. The "qi deficiency" and "qi stagnation" genes based on NEI network were excavated. Conclusion: The genes of qi stagnation pattern relate closely with nerve and endocrine, while those of qi deficiency pattern have a close relationship with the immune system. The intrinsic biological characterizations of TCM patterns can be effectively identified at the level of NEI.

Introduction

In Traditional Chinese Medicine (TCM), diagnosis relies on "syndromes" or "patterns"—subjective classifications of a patient's state, such as Qi deficiency or Qi stagnation. While these concepts have guided treatment for centuries, they lack objective, biological definitions that align with Western molecular medicine.

This research aims to bridge the gap between TCM and modern molecular biology by identifying the intrinsic biological distinctions between these two common patterns. The study focuses on the Neuro-Endocrine-Immune (NEI) network, a modern framework that views the nervous, endocrine, and immune systems as a single, integrated regulatory network. By mapping TCM syndromes onto this molecular network, the researchers propose a more objective and systematic evaluation system for TCM.

Methodology

The researchers employed a multi-step computational pipeline to extract and analyze biological data from scientific literature.

Research workflow for text mining and network analysis of TCM syndromes.

The process involved:

  1. Lexicon Building: Establishing a data dictionary of 2,242 NEI-related genes (sourced from the dbNEI database) and a keyword lexicon for the two TCM syndromes based on standard clinical practice guidelines.
  2. Literature Mining: Using a self-developed tool called PubMedMiner, the team retrieved and processed abstracts from PubMed—over 41,000 for Qi deficiency and nearly 148,000 for Qi stagnation.
  3. Co-occurrence Analysis: The core of the study relies on the assumption that if a gene and a syndrome keyword frequently appear together in the same abstract, they are likely semantically and biologically related.
  4. Network Construction: Specific mathematical formulas were used to calculate the "relevancy degree" (both direct and indirect) between keywords and genes. This allowed for the construction of complex biological networks where "key nodes" were identified as genes with a connectivity degree at least twice the average of the rest of the network.

Results and Molecular Signatures

The study demonstrates that TCM syndromes have distinct molecular signatures within the NEI network.

Qi Stagnation: Neuro-Endocrine Focus

The Qi stagnation network was found to be robust, featuring 317 nodes and 903 edges. Key genes identified include BDNF (Brain-derived neurotrophic factor), INS (Insulin), and CRH (Corticotropin-releasing hormone). These genes are heavily involved in the nervous and endocrine systems, aligning with the TCM view of Qi stagnation often being related to emotional stress and metabolic regulation.

Qi Deficiency: Immune System Focus

Conversely, the Qi deficiency network is smaller and more specialized, with 32 nodes and 118 edges. The dominant genes were MASP1 and various complement components (C2 through C9). This suggests that "Qi" deficiency may be a clinical proxy for the body's innate immune vigor, specifically the complement system.

Clinical Verification

The literature-mined results were verified using clinical data from patients with coronary heart disease (specifically those exhibiting blood stasis combined with either Qi stagnation or Qi deficiency). The findings confirmed that text mining can effectively identify the "intrinsic biological characterizations" of TCM patterns.

Conclusion and Future Directions

The study successfully identified that genes of the qi stagnation pattern relate closely with nerve and endocrine functions, while those of the qi deficiency pattern have a close relationship with the immune system. This confirms that TCM patterns can be effectively characterized at the level of the NEI network.

However, several challenges remain. Many of the identified "key genes" are not routinely tested in standard clinical settings, limiting immediate implementation. Furthermore, the clinical verification was restricted to coronary heart disease; future work should expand this scope to other diseases to ensure the universality of these biological markers. Further pharmacological studies are also required to understand the exact mechanisms by which these genes produce TCM clinical manifestations.

Summary updated

Abstract

Objective: To make intrinsic biological distinctions between qi deficiency pattern and qi stagnation pattern with genes related to NEI. Methods: Establish a data dictionary of NEI-related genes and a keyword lexicon of qi deficiency pattern and qi stagnation pattern. Then retrieve relevant literature on Pubmed database. Obtain compositions of characteristic NEI-related genes of qi deficiency pattern and qi stagnation pattern by using text-mining methodto explore different bioactive materials between the two syndromes. Results: Two kinds of syndromes and their biological networks were constructed. The "qi deficiency" and "qi stagnation" genes based on NEI network were excavated. Conclusion: The genes of qi stagnation pattern relate closely with nerve and endocrine, while those of qi deficiency pattern have a close relationship with the immune system. The intrinsic biological characterizations of TCM patterns can be effectively identified at the level of NEI.
BibTeX
@misc{zhai2017use,
  author = {Xing Zhai and Ping Wang and Youliang Huang},
  title = {{Use Text Mining and Complex Networks to Analyze TCM Syndromes}},
  year = {2017},
  doi = {10.2991/bbe-17.2017.78},
}
Figure