Research TCM Clinical Diabetes Data Mining

Diabetes is a disease, which serious harm to human health and it has a high incidence, longer duration, more complications. In recent years, the incidence of diabetes increased. Chinese medicines have a good effect in improving the symptoms of diabetes and prevent complications. However, the traditional Chinese medicine diagnosis due to its own characteristics, there are subjective and repeatability. Such serious hinder to Chinese medicines academic thought and clinical diagnostic experience exchanges and research. This subject introduce data mining techniques into Chinese medicine clinical data processing, adopting appropriate data mining method, using Kunming Municipal Hospital of Traditional Chinese Medicine clinical data as the data source, try to find out the regularity of the treatment of diabetes, in order to provide a reference for TCM clinical treatment of diabetes.Firstly, we extracted patients’ symptoms, dialectical and prescription information from diabetes electronic medical records, standardized and numerical processing for the existence of the phenomenon such as "one meaning multi-word","description tedious" and "the meaning of the word overlap" in the information, on this basis,use Apriori algorithm to find out the association rules between any two of the "symptoms","dialectical","prescription", dig out the law of the traditional Chinese medicine treatment of diabetes.Secondly, we process "symptoms" information again, deleting the symptoms which are unrelated with the dialectical classification, merging the symptoms which are the same parts of the patients, then use decision tree C4.5algorithm building a diabetes dialectical classification tree based on symptoms.Compareing related traditional Chinese medicine theory to evaluate these rules. The experiments show that using algorithm not only can dig out laws which subject of the Chinese medicine syndrome, but also can play the role of riching Chinese medicine theory.