基于KNN核函数聚类的轮状病毒统计分析

Clustering of Rotavirus Based on KNN- kernel Function

[目的]探讨K最近邻(K-Nearest Neighbors,KNN)核函数聚类方法在腹泻患者血清免疫指标分类诊断中的适用性和临床意义。[方法]利用KNNCLUST算法的原理和步骤用Matlab软件进行编程,对74例腹泻患者的血清免疫指标数据进行聚类分析,揭示KNN-核函数聚类方法在腹泻患者血清免疫指标分类诊断中的适用性和临床意义。[结果]74例患者经聚类分析分成了5类。该分类不仅把腹泻患者分成轮状病毒阴性和阳性两类,而且把患者进一步进行细分,尤其是把3个初期轮状病毒检测阴性但后期证实是阳性的患者聚成一类。[结论]应用基于KNN-核函数的非参数聚类方法,有助于筛选前期轮状病毒感染者,对疾病的早期诊断治疗具有一定临床意义。

[Objective] To discuss application of KNN-kernel clustering methods for diarrhea patients serum immune indexes detection data classification and diagnosis of applicability and clinical significance. [Methods] To reveal the applicability and clinical signnificance of KNN-kernel function clustering method in the diagnosis of serun immune index. In this research, the KNNCLUST algorithm is used to program the serum immune index data of 74 patients with diarrhea by Matlab software. [Results] 74 patients were divided into 5 categories by cluster analysis. The patients with diarrhea were divided into rotavirus negative and positive class, and the patients were further subdivided, especially the three early rotavirus tests were negative but later confirmed positive and were clustered into one group. [Conclusions] This can be seen that the KNN-kernel clustering method is helpful for early screening of rotavirus infection, practical clinical significance on the early treatment of disease.