Classification of Gastric Diseases Based on Tongue and Nasal Features in Traditional Chinese Medicine Using Machine Learning

Utilizing visual diagnosis based on traditional Chinese medicine (TCM) is a crucial direction for intelligent gastric disease analysis. However, existing research has limited the extraction of visual diagnostic features, which impacts classification accuracy. Therefore, this study proposes a gastric disease classification method based on the analysis of TCM tongue and nasal features combined with machine learning. This method is grounded in visual diagnosis, selectively extracting ten parameters encompassing tongue color, texture, edges, and nasal color. Additionally, age, gender, and season of consultation information are incorporated to establish a feature dataset. The feature dataset is then inputted into four common machine learning algorithms to train a gastric disease classification model. Experimental results demonstrate that employing support vector machines as the classifier yields the best performance, with an achieved AUC value of 0.761 (P<0.05), indicating its diagnostic value. Furthermore, the results provide further evidence of the feasibility of utilizing intelligent information from TCM for early screening of gastric diseases in patients, thereby supporting the promotion of the concept of "preventive treatment with TCM."