A Study On Better Models For Association Rule Mining In Chinese Medicine Prescriptions And Their Applications

Objective: In this thesis,we study the association rule mining for traditional Chinese medicine?TCM?prescriptions and their dosses,compare and optimize association rule models in the context of prescriptions for hepatic cirrhosis,and finally demonstrate the mining results through visualization technology.Methods: Through literature review,we have a comprehensive understanding of TCM prescription data mining and association rule mining.Based on the average doses of prescription and pharmacopoeia drug doses,we create two kinds of weighted support.We use the SPSS Modeler,Weka and our own TCM prescription data mining simulation platform to carry out the mining tasks,under the guidance of TCM clinical experts,and effectively improved two association rule minning models.Results: 1.By incorporating the prescription information,we constructed the weighted models:?1?Weight formula based on drug mean dose: w₁=min?dose_cf?herb??dose_jz ?T,herb??2:1,…??2?The Weight Formula of the Effect of Reducing the Efficiency of Drug: w₂=ZXTable.contains?set??2:12.The improved model in the experimental phase relative to the traditional model,generate more frequent itemsets,and less than the previous rules,that is,improved model to improve the excavation results of the degree of conciseness.3.We constructed a TCM prescription data mining simulation platform.4.We show the results of mining,with enhanced readability via data visualization techniques.Conclusion: This paper builds the weighted support by incorporating the traditional Chinese medicine prescription doses,thus improving the traditional association rule algorithms.The results of the two improved models are more meaningful than the traditional models.In summary,dose-weighted association rule mining models developed in this study effectively improves the mining of TCM prescription combinations.