Study On Data Mining Methods And Applications In Traditional Chinese Medicine Analysis

Traditional Chinese medicine is one of China's four quintessence.In the process of development of thousands of years,the formation of the treatment prescription law theory and the selection of drugs unique syndrome differentiation.It has made an important contribution to human health,and it is the precious wealth of the human.With the modernization of the traditional Chinese medicine,its research unceasingly thorough,TCM and multidisciplinary mutual penetration,formed a trend of interdisciplinary research.Data mining can discover the potential meaning of knowledge in a large,nonstandard,rough and incomplete data set.It is the key technology of knowledge discovery,and it has been widely used in various fields.Data mining has also started in the field of traditional Chinese medicine application.Unlike traditional literature research,data mining technology is to find the model and explore the law according to TCM data objectively,without subjective factors.The research of traditional Chinese medicine has shown good prospects,and can open up a new way for the research of Chinese medicine.On the basis of studying the current research situation of data mining in the field of TCM,original data of formulae that release the exterior with pungent-warm herbs was collected.Data was cleaned and standardized by books,monographs and materials,from several aspects such as prescription,drug name,disease data and dosage unit.After data cleaning and standard arrangement,valid data were obtained,and the database was established.After cluster analysis,classification analysis,association rule mining,prediction analysis and other preliminary experimental research,the main methods of data mining are studied,such ask-means algorithm,Apriori algorithm and BP neural network algorithm.The cluster analysis model was established by using k-means algorithm for outlier analysis.When clustered into 2 classes,the numbers of members were 212 and 5,indicating the data of formulae that release the exterior with pungent-warm herbs almost all belong to a category.In addition,rules of association between drugs and symptoms also were mined by using apriori algorithm,and found the corresponding relationship between drugs and diseases.These results of association rule mining conform to the theory and practice of TCM,and can provide assistant guidance for clinical medication.Besides,predictive relationship between drugs and symptoms were explored by using BP neural network model.And it was found that the accuracy rate of determining the symptoms of the disease by prescription drug combination was high,and the accuracy rate of selecting the core single drug through the disease was low.