基于信任模型的中医药方剂相似度计算方法

Similarity Calculation Method of Traditional Chinese Medicine Prescriptions Based on Trust Model

为了解决中医药方剂相似度无法定量化和方剂组成成分互相干扰问题,提出了一种基于信任模型的中医药方剂相似度的计算方法,通过方剂自身交互经验,计算不同方剂直接信任值,同时依靠两种方剂间的推荐信息来获得第三种方剂信任程度,凭借所得信息计算推荐信任值,使用直接信任值与推荐信任值构建综合信任模型,对信任值较高的方剂与其它信任值较高的方剂交互,确定方剂组成成分分布,利用多变量普遍化分布描述交互方剂,得到组成成分分布规范化因子,构建方剂、证型、组成成分关联,并将该关联存放至信任模型内,依靠信任模型对方剂—证型、方剂—组成成分进行相似度计算。仿真结果证明,所提方法能够更为精确识别方剂之间含有的组成成分,并且不会因为成分重复记录较大,导致计算结果出现误差。

There are some defects in the similarity of traditional Chinese medicine prescriptions,such as difficulty in quantification and mutual interference of prescription components.Therefore,this paper proposes a calculation method about the similarity of traditional Chinese medicine prescriptions based on trust model.First of all,according to the interaction experience of prescriptions,the direct trust value of different prescriptions was calculated.Meanwhile,the trust degree of the third prescription was obtained via the recommendation information between the two prescriptions.Secondly,the recommended trust value was calculated according to the obtained information,and the direct trust value and recommended trust value were adopted to construct the comprehensive trust model.Then,the prescription with high trust value interacts with other prescriptions with high trust value,the composition distribution of prescription was determined.Then,the multivariate generalized distribution was utilized to describe the interactive prescription,and the normalization factor of component distribution was obtained.Prescription,syndrome type and component association were founded and stored in the trust model.Finally,the similarity of formula-syndrome type and formula-component was calculated by the trust model.The simulation results show that the proposed method has high accuracy and low error.