The Application Of Association Rules And Clustering Analysis In Compatibility Of Prescription

Traditional Chinese Medicine (TCM) is not only the quintessence of our country, but also a significant part of the human knowledge treasures. TCM has formed its unique theory of "syndrome differentiation and treatment" with a long history. Simultaneously, a large number of books, as a wisdom collection of ancient Chinese doctor, have been accumulated, especially in the prescription records, which are a great wealth of knowledge. TCM prescription is the primary means of treating disease, study of compatibility of prescription is not only a significant part of research for modern TCM, but also the key link that governs the progress of the modernization of TCM. With the development of information, digitalization, modernization of TCM, prescription database has been increasingly improved, and it is a hot issue that how to discover knowledge from the prescription database.With the development of database technology, data mining as a new data processing and analysis techniques came into being, and has been widely used. Data mining is considered as the nontrivial extraction of implicit, previously unknown, and potentially useful information from data. Apply data mining techniques to analyze the data of ancient prescription, reveal the law of compatibility of prescription, this will play a positive promoting role in the research of compatibility of prescription.This thesis researches the compatibility of prescriptions by data mining techniques. Through modeling drug data, designing and implementing the appropriate algorithm, finally, conclusions have been given in this thesis, which are accorded with the theory of TCM. The detailed contents of the thesis are as follows:(1) Clustering analysis is studied to give drugs a reasonable classification, and also give an evidence to identify the status of drug in prescription in this thesis, K-means algorithm is presented to calculate the problem of drugs clustering for the first time, these include drugs clustering under the single disease and the single prescription.(2) Analyze the correlation between drugs by association rules, these include correlation between drugs under the single diseases and multi-diseases. Based on research of special drug pairs, a method to discover general drug pairs is presented in this thesis. Simultaneously, association rules between diseases are studied. In addition, based on parameter sensitivity analysis, a reasonable parameter is designed, and thesis also introduces the concept of interestingness.(3) Design and develop the ancient prescriptions data mining system based on B/S structure and embed the Apriori algorithm and K-means algorithm into the system, complete the corresponding data mining functions.