Research And Application Of Key Algorithms For Data Mining Of Traditional Chinese Medicine Prescriptions

With the deepening of informationization,Traditional Chinese Medicine is the Chinese traditional Medical which has a long history,and accumulats a lot of prescriptions in clinical practice and theory research in thousands of years.Chinese medicine prescriptions related information is becoming more and more.It is an urgent problem to find useful knowledge from these valuable database resources,to analyze the existing TCM prescriptions and to provide the corresponding data support for the research compatibility of Chinese medicine experts.Therefore,based on the characteristics of TCM prescription data,the data mining method was used to analyze the compatibility rules,the dosage of medicinal materials and the information of the corresponding prescription in the treatment of stroke and diabetes so as to screen out the representative decoction.In detail,the main research content contains the following four parts:(1)Data preprocessing and prescription modeling.It is different for prescription name,medicinal name,dosage and other records in existing Chinese medicine prescription classical books,and it brings the uniformity and the integrity for Chinese medicine prescription data.Therefore,the original data collection must be standardized in the construction of TCM prescription database.(2)Frequent pattern mining algorithm based on SO-Sets.We propose SO-Tree(Start Order Coding Tree)data structure and generate the code of SO-Sets based on the optimization of TB-Tree.The elements of SO-Sets only contain start order(or finish order),and greatly improve the efficiency of its space.Based on SO-Sets,we propose the effective algorithm to mine frequent patterns of traditional Chinese medicine prescription.We adopt the property of superset equivalence to reduce the search space.(3)Top-Rank-k frequent patterns minning algorithm based on Weighted Undirected Graph.To reduce the effect of TCM Prescription mining results due to too much or too small input parameter and filter out the l-itemsets and 2-itemsets,and directly dig out the frequent k-itemsets(k? 3),an efficient algorithm,based on Weighted Undirected Graph,is proposed to discover Top-Rank-k frequent patterns of TCM Prescription.With mining the core drug combination meeting the conditions,the new algorithm can consequently rapidly backtrack to the prescription of the frequent item sets of core drugs combination and provide the data of TCM Prescription for Chinese medicine experts.Besides,the compression mechanism of Dynamic Bit Vector is used to.store compressed for the weights of edges in undirected graph in order to improve the efficiency of the algorithm storage space.(4)Maximal weighted frequent itemset mining algorithm based on WIdT-Tree.In view of the medicinal materials in different prescriptions representing different value and significance,the concept of dTidset is defined.On the basis of dTidset,WIT-Tree storage structure is improved and optimized,and an efficient algorithm,based on WIdT-Trees,is proposed to discover maximal weighted frequent patterns.Besides,The Diffset strategy is used to calculate some frequent k-item sets,and the properties of the equality of item sets weighted extended support in order to reduce the computational cost and decrease the cost of search space.After analysis of the law of core drugs combination,the analysis of the core drug combination,the prescription information and the dosage of drugs can provide important data support for the compatibility of stroke and diabetest.It is the great significance to enrich the theory of formulaology on TCM.