基于SVM的心律失常的研究和分析

Research and Analysis of Arrhythmia Based on SVM

心电图是诊断心脏活动的重要工具。本文从心电图的特征选取和SVM的核函数参数以及惩罚因子优化等提高对心律失常的异常波检测识别。心电特征选取利用小波变换提取了时域特征等参数,将二分的SVM运用到心律失常的检测中。SVM中通过高斯核函数参数和惩罚因子的选取来提高分类的精度。通过对MIT-BIH数据库的数据分析和验证,证明所选取的特征以及参数的可以提高对心律失常分类的正确性。

ECG is an important tool in the diagnosis of cardiac activity. This paper discusses ECG feather extraction and kernel function of SVM in improving the detection of arrhythmia. ECG feather extraction uses wavelet transform to extract the time domain feather and other parameters. Two points SVM to the detection of arrhythmia are employed. The accuracy of classification is improved by the selection of the parameters of the Gauss kernel and the penalty factor in SVM. Through data analysis and verification of MIT-BIH database, the feather and parameters of the selected feath-ers prove to be improved.