Research Article
A Multi-classifier Fusion Approach for Capacitive ECG Signal Quality Assessment
@INPROCEEDINGS{10.1007/978-3-031-06368-8_17, author={Zhikun Lie and Yonglin Wu and Guoqiang Zhu and Yang Li and Chen Chen and Wei Chen}, title={A Multi-classifier Fusion Approach for Capacitive ECG Signal Quality Assessment}, proceedings={Wireless Mobile Communication and Healthcare. 10th EAI International Conference, MobiHealth 2021, Virtual Event, November 13--14, 2021, Proceedings}, proceedings_a={MOBIHEALTH}, year={2022}, month={6}, keywords={capacitive ECG Signal quality assessment Support vector machine K-nearest neighbor model Decision tree model}, doi={10.1007/978-3-031-06368-8_17} }- Zhikun Lie
Yonglin Wu
Guoqiang Zhu
Yang Li
Chen Chen
Wei Chen
Year: 2022
A Multi-classifier Fusion Approach for Capacitive ECG Signal Quality Assessment
MOBIHEALTH
Springer
DOI: 10.1007/978-3-031-06368-8_17
Abstract
Capacitive ECG (cECG), as a contactless solution for measuring ECG, has been extensively explored in existing works. However, the signal quality obtained by cECG can abruptly degrade due to body movement. Hence, it substantially increases the challenge in signal quality assessment of cECG. In this paper, a novel multi-classifier fusion approach is proposed to assess the cECG signal quality. It combines three commonly used classifiers namely, support vector machine (SVM), K-nearest neighbor (KNN) model, and decision tree (DT) and fuse these classifiers with a voting mechanism to provide a robust decision. With the proposed approach, the overall accuracy of 98.32% can be achieved in distinguishing the cECG signal quality into three categories, namely clear ECG signal, blurry ECG signal with clear R peaks, and noisy ECG signal. Experimental results exhibit that the proposed method outperforms existing works. The classification accuracy and F1-Score of this method are better than traditional methods. Meanwhile, the proposed method is expected to be integrated with cECG device for practical long-term heart monitoring.
