Research On Syndrome Classification Of Inquiry Diagnosis For Chronic Gastritis In Traditional Chinese Medicine By Extremely Randomized Forest Algorithm

Decision tree is one of the most widely used in machine learning.It is a probabilistic graph model which combined with graph model and probabilistic decision table.Decision tree represents the relationship between observation variables and class in graphical model.This research set up extreme random forest(ERF)algorithm of chronic gastritis in traditional Chinese medicine(TCM)physician and the main work and contributions are show as follows:1.Using the Extreme Random Forest Algorithm to explore the interrelation between symptoms and symptoms,symptoms and syndromes.calculating of the importance of the variables and the size of the mutual information between the syndromes,the distribution of the symptoms of the syndromes was expressed by means of the aggregated graphs,and the diagnostic criteria of TCM syndrome differentiation and syndrome diagnosis of chronic gastritis were analyzed Comparative analysis,the results basically consistent with the theory of TCM syndrome differentiation,and consistent with clinical practice.2.To propose the extreme randomly forests to establish TCM syndrome differentiation model based on feature selection.The leaf nodes of the decision tree can deal with multiple classes,and obtained great classification performance.Compared with other multi-label classification algorithms to verify the effective performance,and the results are tested by statistical analysis methods.It performed outstanding to the most multi-label algorithms and get the average prediction of 83.8%.The model of syndrome classification are analyzed according to the theory of TCM and clinical experience,the results show the model of syndrome classification have great interpretability.