Tongue features enhance early detection of gastric cancer.

A novel non-invasive diagnostic model integrating traditional Chinese medicine tongue features with machine learning shows significant potential for early gastric cancer detection. An analysis of 292 participants revealed distinct tongue characteristics associated with gastric cancer, including bluish-purple and cracked appearances. The Gradient Boosting Decision Tree model demonstrated exceptional performance, achieving an AUC of 0.980. This approach not only highlights the viability of non-invasive methods but also promises to improve early diagnosis in resource-constrained settings.