Autonomic nervous pattern analysis of sleep deprivation. (August 2022)

• Record Type: Journal Article
• Authors: Yang, Xuehong; Kong, Fanmeng; Xiong, Ronglong; Liu, Guangyuan; Wen, Wanhui
• Is Part Of: Biomedical signal processing and control. Volume 77(2022)
• Publication Date: 2022-08
• Subjects: Sleep deprivation -- Autonomic nervous system -- Heart rate variability -- Machine learning; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2022.103735
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Significant inter-group difference of autonomic nervous activities between acute sleep deprivation and normal sleep people. Significant within-group sex difference of autonomic nervous activities of acute sleep deprivation and normal sleep people. Much higher generalization accuracy of the machine learning method to detect sleep deprivation than random guess accuracy. Dominant effect of sleep deprivation on the autonomic nervous system, i.e. the persistent activation of sympathetic nervous system. Most activated sympathetic nervous system by sleep deprivation around 00:00 AM, revealing better intervention of sleep deprivation before 00:00 AM. Abstract: The cumulative long-term effects of sleep deprivation (SD) have been associated with a wide range of deleterious health consequences. In order to detect SD with machine learning methods, this paper analyzed the autonomic nervous patterns (ANP) of the normal sleep (NS) populations and those of the acute sleep deprivation (ASD) individuals in the morning. Besides, we also analyzed the ANP of people before sleep deprivation (BSD) and during sleep deprivation (DSD) at night. Electrocardiogram (ECG) data were acquired from 120 subjects. Twenty-five heart rate variability (HRV) features were extracted for statistical analysis and pattern recognition. Sequential backward selection method was applied to select the critical feature subset of SD pattern recognition. Three classical classifiers were trained with BSD and DSD samples, and their performance were validated and compared with each other, in order to obtain the best parameters of the SD recognition models. The results of statistical analysis showed that the ANP of NS and ASD groups were significantly different. Besides, the ANP of NS and ASD showed significant within-group sex difference. The results of SD pattern recognition showed that, with 3-dimension critical HRV features of the night ECG data, support vector machine classifier obtained validation accuracies of 76.92% for males and 83.02% for females. It is worth noting that, the low-frequency power of RR interval series increases with the increase of SD time, showing that SD leads to persistent activation of sympathetic nervous system.