A Data Mining System for Herbal Formula Compatibility Analysis

摘要

In this paper, to mine herb compatibility and especially how formula emerges by some herb combination, we propose a improved association rule algorithm based on herbs frequency and combination, so all the compatibility relationship is displayed in a tree structure, based on which a new data mining system is introduced to analyze the compatibility of herbal formulae in traditional Chinese medicine. This system is mainly based on a tree-based method, and incorporates functions of data cleaning, data selection, data formatting, formula tree generating and result outputting. Experimental results on datasets of viral myocarditis treatment literature in the past 10 years show that this system could serve as a useful tool for data mining of herbal formula compatibility. By which we can infer clearly the relationship of formulas and their derivation, and also shows that the core of TCM treatment of viral myocarditis are forsythia, honeysuckle and licorice root, as the core drug of treatments., to which more attentions should be paid in future pharmacology research.