复杂系统熵聚堆算法及其在中风病临床的应用

对数据进行非监督聚类是中医临床研究的主流和难点.提出了一种非监督的复杂系统熵聚堆算法.它改进了关联度系数,不但能实现自组织非监督聚类,而且可以实现一个变量分在不同的类里;提出并证明了N-class相关的概念,加快算法的收敛速度.它运用到中风病临床数据中,非监督地提取出了中风病中常见的证候,结果十分符合临床; 参考数据的辨证结果对算法进行了验证,得到算法的敏感度为97.3%, 这验证了算法的有效性.它为中医临床治疗中风病的规范化奠定了数理基础.

Complex Systems Entropy Cluster Algorithm and Its Application in Stroke Clinics

Using unsupervised algorithms to cluster four diagnosis information data is mainstream and difficulty of Traditional Chinese Medicine clinical research.Based on ischemic stroke clinical data collected,an unsupervised complex system entropy cluster algorithm was proposed.The algorithm ameliorates the traditional correlation coefficient,it not only can realize unsupervised cluster,but also can realize that a variable appears in different clusters.An N-class correlation concept was proposed and proved to significantly accelerate the convergence time of the algorithm.The algorithm was applied to the clinical data of stroke,extracting the hackneyed syndromes in the Stroke unsupervisedly.The results accord to clinics significantly.Finally,the supervised part of data was taken into account to validate the algorithm,reaching a sensitivity of 97.3%.In a word,the algorithm paces a benign mathematical and physical base for standardization of healing stroke by Chinese medicine.