关键词: 中医体质, 体质辨识, 机器学习, 深度学习

Abstract: Traditional Chinese medicine (TCM) has thousands of years of experience in preventing diseases, while TCM constitution, as an important part of TCM, is closely related to individual health and thus plays an important role in disease prevention and treatment. In recent years, the rapid development of information technology and artificial intelligence has promoted the widespread application of numerous intelligent technologies in the field of TCM constitution recognition. These technologies not only make the traditional TCM constitution identification process more scientific and systematic, but also provide strong technical support for the modernization of TCM and personalized medicine, aiming to further improve the accuracy and efficiency of TCM constitution identification. In order to promote the research work on intelligent identification of TCM constitution, the research progress of its method is sorted out and summarized. Firstly, a systematic overview of the data analysis-based methods of constitution identification is made from the data level; secondly, the traditional machine learning-based methods of constitution identification are reviewed and discussed, and compared from the perspective of classifiers; lastly, the deep learning-based methods of constitution identification are elaborated and categorized into early neural networks, convolutional neural networks, hybrid networks, and other methods are summarized from the perspective of network architectures. For each of these three methods, they are analyzed according to their research methods and results, comparing their advantages and limitations, and discussing the potential development trends in future research work.

Key words: traditional Chinese medicine constitution, body constitution identification, machine learning, deep learning