基于深度学习的多导睡眠监测仪运行异常状态监测  

Monitoring abnormal operation of polysomnography based on deep learning

作　　者：陈蕾[1] 金剑[2] Chen Lei;Jin Jian(Sleep Medicine Center Of Neurology Department Of Wuhan NO.1 Hospital,Hubei Wuhan,430000;Equipment Department Of Wuhan NO.1 Hospital,Hubei Wuhan,430000)

机构地区：[1]武汉市第一医院神经内科睡眠医学中心,湖北武汉430000 [2]武汉市第一医院设备处,湖北武汉430000

出　　处：《现代科学仪器》2022年第4期234-238,248,共6页Modern Scientific Instruments

摘　　要：多导睡眠监测仪运行异常状态监测由于非稳态离散故障信号的影响,导致监测精度低和监测耗时长,提出基于深度学习的多导睡眠监测仪运行异常状态监测方法。识别多导睡眠检测仪中运行的异常信号,分离存在损伤的非稳态离散故障信号,建立异常状态数据转换通道,根据深度学习算法获取监测仪运行数据特征,完成多导睡眠监测仪运行异常状态监测。实验结果表明,监测方法的监测精度在90%以上,平均监测耗时仅为1.52s,该方法有效提高了监测精度,降低了监测耗时。Due to the influence of unsteady discrete fault signals,the monitoring of abnormal state of polysomnography leads to low monitoring accuracy and long monitoring time.A method for monitoring abnormal state of polysomnography based on deep learning is proposed.Identify abnormal signals running in the polysomnography detector,separate the non-steady discrete fault signals with damage,and establish a data conversion channel for abnormal state;obtain the operating data characteristics of the monitor according to the deep learning algorithm,and complete the abnormal state of the polysomnography monitor.monitor.The experimental results show that the monitoring accuracy of the monitoring method is more than 90%,and the average monitoring time is only 1.52s.This method effectively improves the monitoring accuracy and reduces the monitoring time.

关 键 词：深度学习 多导睡眠监测仪 运行异常状态 离散故障信号 异常信号幅值 小波变换函数 

分 类 号：R197.39[医药卫生—卫生事业管理]