Mining The Core Herbs And Their Combination Rules In Traditional Chinese Medicine Formulae

Theory of Traditional Chinese medicine (TCM) as a relative theory to Western medicine has played an important role in the history of medicine civilization. Traditional Chinese medicine formula as an important medical means not only has complex composition but also has huge amount of data.The massive and complexity of TCM prescription hinders further research and development of traditional Chinese medicine. So mining of core herbs and their combinations not only can reveal the knowledge underly formulae,but also can simplify the formulae which can easy the inheritance and research of TCM theory.For the extraction of core herbs and their combinations, this paper proposes a method including four steps.The first step is the clustering of traditional Chinese medicine formulae (TCMF). This chapter presents an improved k-means algorithm. The experiment proves that the algorithm is better than k-means. Finally, the algorithm is applied to the clustering of TCMF, also has achieved good results.The second step is the discovery of core herbs inTCMF. This chapter presents two methods for herb effect measurement. By experiments, it is confirmed that the two measurements can reasonably find the core herbs.The third step is the measurement of herbs interaction. This chapter proposes an interaction measurement called PMIEDFL, which is based on PMI, formula length and herb pair ED. We confirm that PMIEDFL can reasonably measure the interaction between herbs by experiments.The last step is the discovery of herb combinations. The chapter proposes a method to find the herb combinations based on the communities in herb network.A new node fiteness measurement and an algorithm modified from the algorithm proposed by Andrea Lancichinetti are presented in this chapter for the discovery of herb combinations. Experiments show that this algorithm can found the herb combinations in formulae...