Research of TCM syndromes diagnostic models for chronic gastritis based on multielement mathematical statistical methods

Abstract

In this study, we assessed the large sample population of patients with chronic gastritis based on three methods with supervised learning function, i.e., the regression analysis, BP neural network and Support Vector Machine. On basis of the results, we constructed the diagnostic models to predict the types of Traditional Chinese Medicine (TCM) syndromes of chronic gastritis, and compared the correct rate and applicability of each method. The study showed the correct rate of prediction was as follows: Support Vector Machine ≫ BP neural network ≫ regression analysis, after construction of diagnostic models with three algorithms. We believe, our results could be of great value in exploring the methodology of objectification and standardization of TCM Syndromes.