摘要
目的设计一种在心肺复苏干扰下识别心电节律的算法,以此减少心电节律识别错误率,缩短分析时间。方法采用13个特征值构建神经网络,基于机器学习的方法优化算法,增加心电节律识别的准确率。结果设计算法在干扰下对心电节律的识别有很高的准确率,尤其在VT的识别上,达到100%的准确率,即使在干扰较强(SNR=-12)时,对各种类型节律的辨识准确率也均在95%以上。结论构建的算法在心电节律识别时鲁棒性较强,通过基于受干扰心电信号数据进行的性能评估,证明该算法即使在信号受到特别高水平的干扰时,也能实现室颤和室速节律的准确辨识。
Objective To design an algorithm to identify ECG rhythms under CPR interference as a way to reduce the ECG rhythm identification error rate and shorten the analysis time.Methods 13 eigenvalues were utilized to construct the neural network,and the algorithm was optimized based on machine learning to increase the accuracy of ECG rhythm recognition.Results The designed algorithm had high accuracy rate in the recognition of ECG rhythm under interference,especially 100%accuracy in the recognition of VT,even when the interference was strong(SNR=-12),the recognition accuracy of all types of rhythm was above 95%.Conclusion The proposed algorithm has strong robustness in ECG rhythm recognition.Through performance evaluation based on disturbed ECG signal data,it is proved that the algorithm achieves accurately discrimination of ventricular fibrillation and ventricular tachycardia rhythms even when the signal is subjected to particularly high levels of interference.
作者
余明
袁晶
张广
万宗明
陈锋
YU Ming;YUAN Jing;ZHANG Guang;WAN Zongming;CHEN Feng(Institute of Medical Support Technology,Institute of Systems Engineering,Academy of Military Sciences,Tianjin 300161,China)
出处
《中国医疗设备》
2021年第6期31-34,共4页
China Medical Devices
基金
国家重点研发计划(2017YFC0806406,2017YFC0806404,2017YFC0806402)
天津市重大科技计划(18ZXJMTG00060)。
关键词
机器学习
心肺复苏干扰
心电节律识别算法
machine learning
ventricular fibrillation
ventricular tachycardia.