Research On Data Warehouse For TCM Prescription In Chronic Kidney Disease

The Chronic Kidney Disease(CKD)is a common chronic disease.Using the data warehouse for storing the theme of CKD medical records and displaying the compatibilities of prescription in multiple dimensions makes an advantage on learning experience of CKD diagnosis in Traditional Chinese Medicine(TCM).Note-generation and diagnostic path recommendation based on the medical records would be propitious to the inheritance of TCM.In this thesis,the compatibilities of prescription including nature-flavor-meridian tropism,sovereign-minister-assistant-courier,and seven emotions are studied.Data warehouse is used to show the compatibilities of prescription for CKD,and data mining is used to generate note and recommend the path of diagnosis.Specific works are as follows:1.In the research of identifying the sovereign-minister-assistant-courier,the improved naive bayesian classification is proposed to identify the roles of herbs based on the efficacy of prescription,the relative position of herbs and the relative dose of herbs which average accuracy is 85.9% for solving the problem that the existing methods cannot make full use of herb information and the dichotomy structure cannot fully reveal the composition structure of the prescription.This method is compared with the methods based on definition and fuzzy clustering.In the research of identifying the nature-flavor-meridian tropism,a screening method based on the improved Apriori algorithm is proposed to solve the problem that herb dose cannot reflect the real property,which has the accuracy of 86.9%,65.1%,63.3%,respectively.We also compare this screening method with the methods based on quantification of nature/flavor and the coefficient of meridian tropism,respectively.In the research of identifying the seven emotions,the matrix of herbs is established to calculate herb pairs by mutual information,and the seven emotions are identified based on the multilayer perceptron for sovling the problem that the decision tree cannot be built by herb efficacy,which average accuracy is 66.8%.2.In the design of data warehouse for TCM prescription in CKD,the themes of role,nature,meridian tropism and seven emotions of herbs are established in this thesis to analyze the compatibilities of prescription in multiple dimensions and theme-oriented data storage is set up.Based on the theory of dimension model in data warehouse,we determine the fact,dimension and measurement of themes,then constructe the bus matrix of data warehouse and establishe the fact tables and dimension tables with star-model for each theme.The design of ETL is achieved by ODI,and the design of OLAP is completed by OBIEE.3.In the analysis of data mining,the methods of caculating symptom corresponding pathogen and representative prescription,analysising the increasement of herbs and the efficacy of herb pairs,and analysising the herb law on multiple diagnosises are designed for sovling the problem that existing notes are concluded by doctors manually and cost large amount of manpower.These methods are conducive to generate note automatically.Simultaneously,the recommendation of diagnosis path is realized based on the improved Apriori algorithm.4.The system prototype is achieved by WPF,the Oracle data warehouse designer and Echarts.The function of system contains the design of the themes of role,nature,flavor,meridian tropism and seven emotions in multiple dementions,the report exhibition of themes and the path of diagnosis.