基于SVM的人体心电监测算法研究

Research on Human ECG Monitoring Algorithm Based on SVM

摘要: 针对传统心电监测算法噪声鲁棒性差、分类精度低的问题，本文提出一种基于支持向量机(Support Vector Machine, SVM)的心律失常识别算法。首先对麻省理工学院–贝斯以色列医院心律失常数据库(MIT-BIH Arrhythmia Database)的心电信号进行小波去噪与基线校正预处理；然后提取时域、频域特征并优化维度；最后通过网格搜索优化SVM参数，构建多分类模型。实验结果表明，所提算法在准确率、灵敏度、特异性上均优于传统方法，可有效实现心电异常自动识别，为便携式心电监测与远程医疗系统提供算法支撑。

Abstract: Aiming at the problems of poor noise robustness and low classification accuracy of traditional ECG monitoring algorithms, this paper proposes an arrhythmia recognition algorithm based on Support Vector Machine (SVM). Firstly, wavelet denoising and baseline correction preprocessing are performed on the ECG signals from the MIT-BIH Arrhythmia Database; then, time-domain and frequency-domain features are extracted and their dimensions are optimized; finally, SVM parameters are optimized through grid search to construct a multi-classification model. Experimental results show that the proposed algorithm is superior to traditional methods in accuracy, sensitivity, and specificity, and can effectively realize automatic recognition of ECG abnormalities, providing algorithmic support for portable ECG monitoring and telemedicine systems.