常见机器学习方法在中医诊断领域的应用述评

Review of Application of Machine Learning Methods in the Field of Diagnostics of Traditional Chinese Medicine

对近年来数据挖掘技术中的常见机器学习方法应用于中医诊断的概况进行述评。近年来应用于中医诊断的常见机器学习方法有支持向量机、决策树、K近邻算法、人工神经网络、贝叶斯网络、AdaBoost算法等,这些方法各有其优势与不足。今后在促进机器学习运用于中医诊断研究领域时,需基于中医证的多维性、非线性和复杂性的特点,针对中医诊疗中的各项问题制定统一的标准并规范收集数据,选用合适的机器学习方法,寻找中医大数据下的隐藏知识和规律,以推动中医诊断客观化、规范化。

The status of the application of common machine learning methods of the data mining technology in the field of diagnostics of traditional Chinese medicine(TCM)in recent years was reviewed.The common machine learning methods adopted in the field of the research of diagnostics of TCM in recent years included support vector machine,decision tree,K-nearest neighbor algorithm,artificial neural network,Bayesian network,and AdaBoost algorithm,and each of them has its advantages and insufficiency.It is suggested that the future research of diagnostics of TCM with the common machine learning methods should be based on the characteristics of multidimension,non-linearity and complexity of TCM syndrome,collecting the data of diagnostics of TCM should follow the unified criteria,the proper machine learning methods should be chosen for unveiling the hidden knowledge and regularity of TCM diagnostics,so as to promote the objectivity and standardization of TCM diagnostics.