Methodology study of classification algorithm in traditional Chinese medicine syndrome study

Abstract

Study of traditional Chinese medicine (TCM) syndromes is a key to the research of TCM modernization, and the core is the classification and diagnostic criteria of syndromes. The purpose of this article is to review the usage of classification algorithms of data mining in TCM syndrome researches, and comprehensively analyze the main features of algorithms and their applications. The appropriate classification algorithm should be chosen according to different research purposes. Rough sets and cluster analysis are suitable for exploratory research without requiring a prior knowledge. Fuzzy sets theory, neural networks and decision tree are suitable for syndrome diagnostic criteria research when the classification goal is clear, because they require a prior knowledge. Among them, fuzzy sets theory could be used in combination with other classification algorithms. Thus, some new methods such as fuzzy clustering, fuzzy rough sets or fuzzy decision tree might be more suitable for TCM algorithm classification research. It is suggested that some novel classification algorithms need to be developed to fit the condition of TCM syndrome, based on the interdisciplinary theories and technologies.