Abstract: Objective To explore the clinical strategy and method of the preliminary discrimination model of prognosis and survival in the treatment of advanced esophageal cancer with Traditional Chinese Medicine(TCM). Methods Univariate test was performed on multiple factors affecting the prognosis of advanced esophageal cancer, and variables with statistical significance were included. Non-parametric discriminant analysis was used to establish a survival prediction model of traditional Chinese medicine treatment for advanced esophageal cancer. Totally 72 patients with advanced esophageal cancer were included in the study, which were divided into the dominant group, the intermediate group and the disadvantaged group based on the overall survival. Results Univariate analysis showed that location of onset (including upper esophagus), location of onset (including middle esophagus), differentiation degree of cancer cells, metastasis, hot food, smoking, drinking, symptom assessment after TCM treatment, emotion assessment after TCM treatment and lung metastasis showed statistical significance (α=0.1). Conclusions After incorporating the elements of TCM characteristic syndromes, the prediction model is established through 7/3 verification (70% samples as training set, 30% samples as verification set), and the prediction accuracy is 56%.

Key words: advanced esophageal cancer, Traditional Chinese Medicine, the model of prognosis