基于自注意力机制和单导联心电信号的自动睡眠分期算法#br#

计算机与现代化 ›› 2022, Vol. 0 ›› Issue (12): 50-59.

基于自注意力机制和单导联心电信号的自动睡眠分期算法#br#

(1.中国科学院空天信息创新研究院,北京100190;2.中国科学院大学电子电气与通信工程学院,北京100049)

出版日期:2023-01-04 发布日期:2023-01-04
作者简介:栗伟松(1998—),女，河南上蔡人，硕士研究生，研究方向：医疗健康物联网技术，E-mail: liweisong19@mails.ucas.ac.cn; 汤敏芳(1996—),女,博士研究生,研究方向:医疗健康物联网技术,E-mail: 1293227181@qq.com; 何征岭(1993—),男,博士研究生,研究方向:医疗健康物联网技术,E-mail: hezhengling15@mails.ucas.ac.cn; 王鹏(1995—),男,硕士,研究方向:医疗健康物联网技术,E-mail: wangpeng01@aircas.ac.cn; 杜利东(1981—),男,副研究员,硕士生导师,研究方向:微纳制造技术,E-mail: lddu@mail.ie.ac.cn; 方震(1976—),男,研究员,博士生导师,研究方向:医疗健康物联网技术,E-mail: zf@mail.ie.ac.cn; 通信作者：陈贤祥(1979—),男,副研究员,硕士生导师,研究方向:医疗健康物联网技术,E-mail: xxchen@mail.ie.ac.cn。
基金资助:
国家重点研发计划项目(2020YFC2003703)

Automatic Sleep Staging Algorithm Based on Self-attention Mechanism and Single Lead ECG

(1.Aerospace Information Research Institution, Chinese Academy of Sciences, Beijing 100190, China； 2.School of
Electronic, Electrical and Communication Engineering, University of Chinese Academy of Sciences, Beijing 100049, China)

Online:2023-01-04 Published:2023-01-04

摘要/Abstract

摘要： 基于手工标记或传统机器学习方法实现睡眠分期过程复杂且效率低下，深度神经网络因其强大的提取复杂特征的能力改善了睡眠分期结果，但仍存在忽略片段内部信息相关性的问题。针对此问题，本文提出一种基于自注意力机制和单导联心电信号的自动睡眠分期算法，利用卷积模块、双向门控循环单元及自注意力机制实现特征自动提取与分类。在开源睡眠心脏健康研究数据库(SHHS1、SHHS2)、动脉粥样硬化的多民族研究数据库(MESA)和美国麻省理工的多导睡眠数据库(MITBPD)中分别选取1000、1000、1000和16名受试者的单导联心电信号数据进行训练和测试，得到模型睡眠四分类(觉醒、快速眼动期、浅睡眠和深睡眠)结果，分类准确率分别达到75.77%(kappa＝0.63)、81.01%(kappa＝066)、82.79%(kappa＝0.71)和76.22%(kappa＝0.58)，优于基于传统机器学习算法的睡眠分期结果，验证了提出模型的有效性。

关键词: 自动睡眠分期, 单导联心电信号, 卷积网络, 门控循环单元, 自注意力机制

Abstract: The realization of sleep staging based on manual labeling or traditional machine learning methods is complex and inefficient. Deep neural network improves the results of sleep staging because of its powerful ability to extract complex features, but there are still some problems, such as ignoring the correlation of internal information. To solve this problem, this paper proposes an automatic sleep staging algorithm based on self-attention mechanism and single lead ECG signal, realizing feature extraction and classification automatically by using convolution module, bidirectional gated recurrent unit and self-attention mechanism. In the open database Sleep Heart Health Study database (SHHS1, SHHS2), Multi-Ethnic Study of Atherosclerosis database (MESA) and MIT-BIH Polysomnographic database (MITBPD), the single lead ECG data of 1000, 1000, 1000 and 16 subjects are randomly selected for training and testing. The experimental results show that the accuracy of the four sleep classifications (wake, rapid eye movement, light sleep and deep sleep) of the model is 75.77% (kappa＝0.63), 81.01% (kappa =0.66), 8279% (kappa＝0.71) and 76.22% (kappa＝0.58) respectively, which is better than the sleep staging results based on the traditional machine learning algorithms, verifying the validity of the model.

Key words: automatic sleep staging, single lead electrocardiogram, convolution network, gated recurrent unit, self-attention mechanism

栗伟松, 汤敏芳, 何征岭, 王鹏, 杜利东, 方震, 陈贤祥. 基于自注意力机制和单导联心电信号的自动睡眠分期算法#br#[J]. 计算机与现代化, 2022, 0(12): 50-59.

LI Wei-song, TANG Min-fang, HE Zheng-ling, WANG Peng, DU Li-dong, FANG Zhen, CHEN Xian-xiang. Automatic Sleep Staging Algorithm Based on Self-attention Mechanism and Single Lead ECG[J]. Computer and Modernization, 2022, 0(12): 50-59.

参考文献

［1］ PARKES K R. Age and work environment characteristics in relation to sleep: Additive, interactive and curvilinear effects［J］. Applied Ergonomics, 2016,54(6):41-50.
［2］ IBER C, ANCOLI-ISRAEL S, CHESSON A L, et al. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications［M］. Westchester: American Academy of Sleep Medicine, 2007.
［3］李杰. 非脑电睡眠监测系统和算法研究［D］. 杭州:浙江大学, 2016.
［4］ ROEBUCK A, MONASTERIO V, GEDERI E, et al. A review of signals used in sleep analysis［J］. Physiological Measurement, 2013,35(1). DOI:10.1088/0967-3334/35/1/R1.
［5］ EBRAHIMI F, SETAREHDAN S, NAZERAN H. Automatic sleep staging by simultaneous analysis of ECG and respiratory signals in long epochs［J］. Biomedical Signal Processing and Control, 2015,18:69-79.
［6］ FONSECA P, LONG X, RADHA M, et al. Sleep stage classification with ECG and respiratory effort［J］. Physiological Measurement, 2015,36(10). DOI:10.1088/0967-3334/36/10/2027.
［7］ UTOMO O K, SURANTHA N, ISA S M, et al. Automaticsleep stage classification using weighted ELM and PSO on imbalanced data from single lead ECG［J］. Procedia Computer Science, 2019,157(C):321-328.
［8］ REDMOND S J, CHAZAL P D, O′BRIEN C, et al. Sleep staging using cardiorespiratory signals［J］. Somnologie-Schlafforschung und Schlafmedizin, 2007,11(4):245-256.
［9］ YCELBA U, YCELBA C, TEZEL G, et al. Automatic sleep staging based on SVD, VMD, HHT and morphological features of single-lead ECG signal［J］. Expert Systems with Applications: An International Journal, 2018,102(C):193-206.
［10］王爱波. 基于隐马尔可夫模型的心率变异性分析在睡眠分期中的应用［D］. 成都:电子科技大学, 2019.
［11］冯静达,焦学军,李启杰,等. 基于心率和呼吸特征结合的睡眠分期研究［J］. 航天医学与医学工程, 2020,33(2):152-158.
［12］JUNG D W, HWANG S H, LEE Y J, et al. Apnea-hypopnea index prediction using electrocardiogram acquired during the sleep-onset period［J］. IEEE Transactions on Biomedical Engineering, 2017,64(2):295-301.〖HJ0.24mm〗
［13］NGUYEN Q N T, BUI P N, LE T Q. In vivo comparison of sleep stage scoring between commercialized wearable devices and polysomnography system［C］// 6th International Conference on the Development of Biomedical Engineering in Vietnam: BME6. 2018:793-800.
［14］LI Q, LI Q, LIU C, et al. Deep learning in the cross-time-frequency domain for sleep staging from a single lead electrocardiogram［J］. Physiological Measurement, 2018,39(12). DOI:10.1088/1361-6579/aaf339.
［15］FRANTZIDIS C A, NDAY C M, CHRISKOS P, et al. A review on current trends in automatic sleep staging through bio-signal recordings and future challenges［J］. Sleep Medicine Reviews, 2020,55(5). DOI:10.1016/j.smrv.2020.101377.
［16］ZHU T Q, LUO W, YU F. Convolution- and attention-based neural network for automated sleep stage classification［J］. International Journal of Environmental Research and Public Health, 2020,17(11). DOI:10.3390/ijerph17114152.
［17］张兴华. 基于心电信号和深度神经网络的睡眠分期研究［D］. 天津:天津工业大学, 2018.
［18］WEI R, ZHANG X, WANG J, et al. The research of sleep staging based on single-lead electrocardiogram and deep neural network［J］. Biomedical Engineering Letters, 2017,8(1). DOI:10.1007/s13534-017-0044-1.
［19］RADHA M, FONSECA P, ROSS M, et al. LSTM knowledge transfer for HRV-based sleep staging［J］. arXiv preprint arXiv:1809.06221, 2018.
［20］VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need［C］// Proceedings of the 31st International Conference on Neural Information Processing Systems. 2017:6000-6010.
［21］ZHANG G, CUI L, MUELLER R, et al. Thenational sleep research resource: Towards a sleep data commons［J］. Journal of the American Medical Informatics Association, 2018,25(12). DOI:10.1093/jamia/ocy064.
［22］QUAN S F, HOWARD B V, IBER C, et al. The sleep heart health study: Design, rationale, and methods［J］. Sleep, 1997,20(12):1077-1085.
［23］CHEN X, WANG R, ZEE P, et al. Racial/Ethnic differences in sleep disturbances: The multi-ethnic study of atherosclerosis (MESA)［J］. Sleep, 2014,38(6). DOI:10.5665/sleep.4732.〖HJ0.2mm〗
［24］GOLDBERGER A L, AMARAL L A N, GLASS L, et al.PhysioBank, physioToolkit, and physioNet: Components of a new research resource for complex physiologic signals［J］. Circulation, 2000,101(23):215-220. DOI:10.1161/01.CIR.101.23.e215.
［25］周志华. 机器学习［M］. 北京:清华大学出版社, 2016:31-35.
［26］HAMILTON P S, TOMPKINS W J. Quantitative investigation of QRS detection rules using the MIT/BIH arrhythmia database［J］. IEEE Transactions on Biomedical Engineering, 1986,33(12):1157-1165.
［27］XIAO M, YAN H, SONG J, et al. Sleep stages classification based on heart rate variability and random forest［J］. Biomedical Signal Processing and Control, 2013,8(6):624-633.
［28］郑捷文,张悦舟,兰珂,等. 基于心率变异性分析的睡眠分期算法研究和验证［J］. 中国生物医学工程学报, 2020,39(4):432-439.
［29］YILMAZ B, ASYALI M H, ARIKAN E, et al. Sleep stage and obstructive apneaic epoch classification using single-lead ECG［J］. BioMedical Engineering OnLine, 2010,9(1). DOI:10.1186/1475-925X-9-39.
［30］WILLEMEN T, DEUN D V, VERHAERT V, et al. Automatic sleep stage classification based on easy to register signals as a validation tool for ergonomic steering in smart bedding systems［J］. Work, 2012,41(S1):1985-1989.

[1]	何思达, 陈平华. 基于意图的轻量级自注意力序列推荐模型[J]. 计算机与现代化, 2024, 0(12): 1-9.
[2]	张晓东1, 白广芝1, 李敏1, 李昊洋2. 基于经验小波变换的油气井产量预测模型 [J]. 计算机与现代化, 2024, 0(12): 53-58.
[3]	李桃迎, 李蒙, 武梦乔. 基于异构时空图卷积网络的出租车客流预测[J]. 计算机与现代化, 2024, 0(11): 13-18.
[4]	祁贤, 刘大铭, 常佳鑫. 基于改进自注意力机制的多视图三维重建[J]. 计算机与现代化, 2024, 0(11): 106-112.
[5]	张惠楠1, 张强1, 孙红霞2. 基于改进时序胶囊网络的油藏生产动态分析模型[J]. 计算机与现代化, 2024, 0(09): 15-19.
[6]	候聪颖, 杨文清, 王召, 程聪. 基于时频自注意力残差时序卷积网络的语音增强[J]. 计算机与现代化, 2024, 0(09): 20-24.
[7]	杨绍祖1, 2, 王海程1, 2, 吴金雅1, 2, 马纪颖1, 2. 基于STAGCN-Informer时空组合模型的风电功率预测方法[J]. 计算机与现代化, 2024, 0(07): 13-20.
[8]	游嘉靖1, 2, 何月顺1, 何璘琳1, 钟海龙1, 2. 基于AHP-CNN的加密流量分类方法[J]. 计算机与现代化, 2024, 0(04): 83-87.
[9]	吴甜甜, 李延恺, 刘阳. 基于多尺度频率注意力的多阶段去雨算法[J]. 计算机与现代化, 2024, 0(02): 50-55.
[10]	李亚平, 王军防, 余红梅, 窦一民, 肖媛, 田继林. Regformer：基于稀疏注意力的输油管道水力压降预测方法[J]. 计算机与现代化, 2024, 0(01): 59-66.
[11]	谢仕斌, 刘梦赤, 唐诗琪, 周瑞平, . 基于多特征提取的时间卷积知识追踪模型[J]. 计算机与现代化, 2023, 0(07): 25-29.
[12]	苏金库, 桂智明. 基于时空特征的短时出租车流量预测[J]. 计算机与现代化, 2023, 0(05): 32-38.
[13]	杨骏, 王劲林, 倪宏, 盛益强, . 工控网络异常检测中基于灵敏度的动态迁移算法[J]. 计算机与现代化, 2023, 0(05): 46-51.
[14]	夏义春, 李汪根, 李豆豆, 葛英奎, 王志格. 结合注意力机制和图神经网络的CTR预估模型[J]. 计算机与现代化, 2023, 0(03): 29-37.
[15]	赵婧华, 杨秋翔. 基于AU-GCN与注意力机制的微表情识别[J]. 计算机与现代化, 2023, 0(03): 48-53.