Key Features
Data Browser
Browse coordinates of various data types (symptoms, formulas, herbs, compounds, drugs, or targets) in a visualized TCM Embedding Space.
TCM Representation
Convert custom symptom patterns or herb combinations into quantitative coordinates in the TCM Embedding Space.
Generation
Based on symptom-herb mapping patterns learned by the model, this tool infers recommended herb combinations from input symptom patterns, or inversely infers the primary symptom patterns from herb combinations.
Relation Exploration
Search related items that are nearest to the input coordinate (a disease or a custom coordinate) within the TCM Embedding Space.
Analysis
Analyze TCM formula attention, outcome associations, indications, practice patterns, and visualize embeddings.
About the Model
Research Overview
This work introduces a Transformer-based foundation model for Traditional Chinese Medicine that constructs a unified TCM Embedding Space (TCM-ES). The model bridges traditional Chinese medicine with modern biomedicine by mapping TCM entities and biomedical concepts into the same embedding space.
Key Innovations
- Unified TCM Embedding Space (TCM-ES) for symptom patterns and herbal therapies
- Transformer autoencoders with contrastive learning for TCM entity relationships
- Mapping of modern biomedical entities (diseases, compounds, targets) to TCM-ES
- Interpretable explorations with attention mechanisms and embedding visualization
TCM Embedding Space Visualization
Biomedical Entity Integration