Traditional Chinese medicine constitution intelligent identification system based on graph search mode and deep learning algorithm

• Xing Zhai,
• Aiqing Han,
• Yuxuan Jia,
• Lin Zhang,
• Fan Yu,
• Ji Wang,
• Shujuan Hou,
• Yan Tang,

Vol. 19, No. 7, July 31, 2025
10.3837/tiis.2025.07.004, Download Paper (Free):

• Traditional Chinese medicine
• constitution identification
• Artificial intelligence
• graph search mode
• Deep Learning

Abstract

This study integrates intelligent technologies like knowledge graphs and deep learning-based image recognition with traditional Chinese medicine (TCM) thinking to enhance TCM with intelligence. It proposes an AI-based TCM constitution identification system that combines "inquiry diagnosis" (symptom diagnosis) and "inspection diagnosis" (image recognition). The identification steps of the TCM constitution intelligent identification system include two stages: symptom identification and image identification. If the results from both methods are consistent, the results are directly output; if the results are not completely consistent, the voting method is used for comprehensive judgment. It uses semantic analysis and graph queries on a TCM constitution knowledge graph for symptom identification, achieving good results. On the basis of deep learning and computer vision technology, ResNet34 and EfficientNetB1 models were used on the dataset of calm constitution and dampness heat constitution, and the face and tongue image constitution recognition models were trained. Both models exhibited high accuracy. The model EfficientNetB1 showed the best performance, with recognition accuracies of face and tongue images at 80.2% and 95.6%, respectively. The accuracy of the two constitutions could be well determined. Prototype testing and user experience studies show that the system achieves intelligent constitution recognition, assisting users in quickly diagnosing their constitution types. This approach provides a practical solution and lays the foundation for developing clinically applicable systems.

Statistics

Cite this article