Clinical study on data mining of TCM Zheng in Chronic Hepatitis B

Abstract

Purpose of the study. Through the study of the value contributed by symptoms to syndrome diagnosis of Chronic Hepatitis B (CHB), that is diagnosis weight, constructing symptom-syndrome scale of CHB, and realizing quantitative diagnosis from symptom to syndrome. Methods. Collecting the information from four diagnostic methods and diagnostic information of patients who were confirmed of Chronic Hepatitis B from November 2009 to April 2010 from 3 hospitals, the researchers construct a clinical record data base of Chronic Hepatitis B, which is used as the dataset for machine learning. Then association rule mining algorithm is adopted to analyze the data automatically. The confidence of the association rule between symptom and syndrome is taken as the weight of symptom contributed to syndrome diagnosis. All the weights of association rule are collected as symptom-diagnostic scale. Then a test data set is used to evaluate the accuracy of the diagnostic scale. Results. (1) Damp-Heat in the Liver and Gallbladder, Yin Deficiency of Liver and Kidney, Liver Qi Stagnation and Spleen Deficiency are three major syndromes of CHB. (2) Spider nevus, white tongue, moderate pulse and soft pulse are of decisive significance to the diagnosis of Liver Qi Stagnation and Spleen Deficiency. Scaly dry skin, heavy sensation in the limbs and body and short pulse have decisive significance to the diagnosis of Yin Deficiency of Liver and Kidney. Three symptoms are of decisive significance to the diagnosis of Damp-Heat in the Liver and Gallbladder, which are spotted tongue, withered tongue and long pulse. (3) 327 rules are achieved. 91 symptoms can be found in three syndromes of the patients. 9 symptoms can be found in the syndromes of DampHeat in the Liver and Gallbladder, Yin Deficiency of Liver and Kidney. 8 symptoms can be seen in Damp-Heat in the Liver and Gallbladder, Liver Qi Stagnation and Spleen Deficiency. 3 symptoms are seen in Yin Deficiency of Liver and Kidney, Liver Qi Stagnation and Spleen Deficiency. The value that every symptom contributes to the syndrome can be regarded as the differential diagnostic evidence of different syndromes. 60 cases are taken as test data set to evaluate the diagnostic scale and the accuracy reaches 83.3%. Conclusion. The diagnostic scale has a good predictive ability. Quantitative diagnosis from symptoms to syndromes can be realized by the confidence of the association rule between symptom and syndrome. The application of this method in other diseases' differentiation is up to be studied.