Challenges for AI methods in Traditional Chinese Medicine

Keywords: AI, Traditional Chinese Medicine, knowledge presentation, LLM, data integration, physiology, databases, omics

TL;DR: We discuss AI tools and methods for Traditional Chinese Medicine, existing databases and medical knowledge presentation

Abstract:

Applications of AI (Artificial Intelligence) in fundamental medicine greatly vary from human genetics to clinical testing, from protein structure prediction studies to electronic health records. Integrative approach to human health relies on traditional approaches of Oriental Medicine having own system of knowledge presentation, symptoms and diagnostics conceptions.
Here we review integrative medical approaches using classical and traditional healthcare methods knowledge from point of view of AI. AI tools are advancing Traditional Chinese Medicine (TCM) by integrating ancient practices with modern data analysis, particularly in diagnostics, drug discovery, and data integration. These tools leverage machine learning, databases, and large language models (LLMs) to handle TCM's complexity, including herbal formulations and physiological modeling. AI applications in TCM include diagnostic systems like Canonai Medical's AI meridian diagnostic tool, which analyzes electrical resistance from 80 acupoints for objective diagnostics. Machine learning aids in tongue image analysis (that is canonical diagnostics for TCM), pulse diagnosis, and syndrome differentiation, improving accuracy over traditional methods. Traditional herbal formulas for TCM are actively studied using omics technologies. AI combines network pharmacology with multi-omics (genomics, proteomics, metabolomics) to decode polypharmacological mechanisms of herbal components, screen active compounds and predict targets for diseases like cancer or inflammation. Note HERB 2.0 pharmacotranscriptomics datasets map herbal effects to gene expression profiles, identifying similarities for drug repositioning.
AI models physiology via systems theory frameworks for holistic TCM mechanisms and LLMs for diagnosis simulation, literature mining, and prescription generation from ancient texts and cases (in Chinese) LLMs extract insights from TCM records, building knowledge graphs for clinical decisions. Several specific AI tools enhance TCM diagnosis by automating and objectifying traditional methods like meridian analysis, tongue inspection, and pulse reading. We note traditional medicine approaches in Russia and Asian countries not yet formalized in computer databases, discuss current trends if data analysis and medical knowledge representation in AI.

Submission Number: 68