The Application And Research Of Association Analysis In Traditional Chinese Medicine Data Mining

Traditional Chinese Medicine (TCM) is a significant part of the human knowledge thesaurus. TCM has long history of thousands of years and has formed its unique theory of "syndrome differentiation and treatment". However, the application of computer technology in analyzing and managing TCM information is simple and experiential all the times. It is difficult to provide logical explanation for the diagnosis and treatment process of TCM.Data mining is the nontrivial process of identifying valid, novel, potentially useful, and ultimately understandable patterns in data. Association analysis is an important aspect of data mining study. Mining of association rules emphasizes particularly on ascertaining the relationship of data in different regions, and finding out the dependent relation among multi-region. The mining method of association rules could be used to analyze the data of TCM diagnosis and treatment, and to educe the potential knowledge in them.This paper is based on the project of "Knowledge Discovery in Studying Diagnosis and Treatment Pattern and Integrated Rx for Individual Apoplexy Cases". It uses the data of Management Information System (MIS), which has been built to record the information of apoplexy diagnosis and treatment for 3 years, to obtain the discipline of TCM diagnosis and treatment for apoplexy. We can establish a feasible TCM deduction knowledge system according to the results.The contentd of this paper are mainly focused on the following aspects:(1) After analyzing the background of the project, I determine the targets of the study, and put forward 4 problems that should be solved.(2) The paper summarizes the data mining technology at the beginning. The emphasis of the paper is to research the association mining technology. This paper recommends the processes of mining frequent items based on the Apriori algorithm and FP-growth algorithm in detail. Since the knowledge discovery in TCM diagnosis and treatment is a complex problem of multi-level, multi-dimension, and quantitative association mining, this paper introduces the particular procedures to deal with multi-level association analysis, multi-dimension association analysis, and quantitative association analysis respectively. (3) A series of data pretreatment methods are put forward in accordance with the data character. These methods including the data cleaning, integration, transformation, and reduction techniques standardize the TCM original data.(4) In the processes of the study, this paper finishes the MATLAB program of data pretreatment and association analysis based on the Apriori algorithm and the displaying of the association rules. Moreover, considering the problem of overmany frequent items, this paper promotes the FP-growth to Key-item Extraction in Frequent Pattern Growth (KEFP-growth) algorithm to mine the meaningful frequent items. It finishes the VC++ program of this algorithm, and achieves a better mining efficiency. These programs make it easy to make out many association rules which TCM experts are satisfied with.To a considerable extent, this paper provides an efficient way to do the further study in TCM knowledge discovery.