Symptom-herb knowledge discovery based on attribute partial ordered structure diagrams

Abstract

As a medical system with a long history, traditional Chinese medicine (TCM) owns extremely rich and valuable experience and knowledge. In this paper, the granulation-based attribute partial ordered structure diagram (APOSD) is proposed to achieve knowledge discovery for spleen Yang deficiency syndrome in TCM. APOSD is discussed with the perspective of granular computing. Each layer of APOSD is viewed as a level of granularity and the nodes in APOSD layers are viewed as information granules. Granulation based generation method of APOSD reduces the complexity of the algorithm and can extract more comprehensive information from multi-level and multi-scale. The proposed method is testified through analysis of prescriptions of spleen Yang deficiency syndrome screening from the comprehensive database of TCM. Symptom patterns, import herbs and their relationship are extracted from different layers and clusters in generated APOSD. Compared with commonly used Apriori association rule analysis algorithm, the method proposed in this paper can not only extract herbs with high confidence but also extract herbs with specific effect, and the association relationships can also be shown on the generated APOSD.