Automatic sleep classification using adaptive segmentation reveals an increased number of rapid eye movement sleep transitions

Henriette Koch, Poul Jennum, Julie Anja Engelhard Christensen*

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

The reference standard for sleep classification uses manual scoring of polysomnography with fixed 30-s epochs. This limits the analysis of sleep pattern, structure and, consequently, detailed association with other physiologic processes. We aimed to improve the details of sleep evaluation by developing a data-driven method that objectively classifies sleep in smaller time intervals. Two adaptive segmentation methods using 3, 10 and 30-s windows were compared. One electroencephalographic (EEG) channel was used to segment into quasi-stationary segments and each segment was classified using a multinomial logistic regression model. Classification features described the power in the clinical frequency bands of three EEG channels and an electrooculographic (EOG) anticorrelation measure for each segment. The models were optimised using 19 healthy control subjects and validated on 18 healthy control subjects. The models obtained overall accuracies of 0.71 ± 0.09, 0.74 ± 0.09 and 0.76 ± 0.08 on the validation data. However, the models allowed a more dynamic sleep, which challenged a true validation against manually scored hypnograms with fixed epochs. The automated classifications indicated an increased number of stage transitions and shorter sleep bouts using models with smaller window size compared with the hypnograms. An increased number of transitions from rapid eye movement (REM) sleep was likewise expressed in the model using 30-s windows, indicating that REM sleep has more fluctuations than captured by today's standard. The models developed are generally applicable and may contribute to concise sleep structure evaluation, research in sleep control and improved understanding of sleep and sleep disorders. The models could also contribute to objective measuring of sleep stability.
Original languageEnglish
Article numbere12780
JournalJournal of Sleep Research
Volume27
Number of pages12
ISSN1365-2869
DOIs
Publication statusPublished - 2018

Keywords

  • REM sleep stability
  • Automatic classification
  • Data-driven segmentation
  • Electroencephalography
  • Sleep stage switching

Cite this

@article{45600663d8da41a99cc1ae01d5bd97c5,
title = "Automatic sleep classification using adaptive segmentation reveals an increased number of rapid eye movement sleep transitions",
abstract = "The reference standard for sleep classification uses manual scoring of polysomnography with fixed 30-s epochs. This limits the analysis of sleep pattern, structure and, consequently, detailed association with other physiologic processes. We aimed to improve the details of sleep evaluation by developing a data-driven method that objectively classifies sleep in smaller time intervals. Two adaptive segmentation methods using 3, 10 and 30-s windows were compared. One electroencephalographic (EEG) channel was used to segment into quasi-stationary segments and each segment was classified using a multinomial logistic regression model. Classification features described the power in the clinical frequency bands of three EEG channels and an electrooculographic (EOG) anticorrelation measure for each segment. The models were optimised using 19 healthy control subjects and validated on 18 healthy control subjects. The models obtained overall accuracies of 0.71 ± 0.09, 0.74 ± 0.09 and 0.76 ± 0.08 on the validation data. However, the models allowed a more dynamic sleep, which challenged a true validation against manually scored hypnograms with fixed epochs. The automated classifications indicated an increased number of stage transitions and shorter sleep bouts using models with smaller window size compared with the hypnograms. An increased number of transitions from rapid eye movement (REM) sleep was likewise expressed in the model using 30-s windows, indicating that REM sleep has more fluctuations than captured by today's standard. The models developed are generally applicable and may contribute to concise sleep structure evaluation, research in sleep control and improved understanding of sleep and sleep disorders. The models could also contribute to objective measuring of sleep stability.",
keywords = "REM sleep stability, Automatic classification, Data-driven segmentation, Electroencephalography, Sleep stage switching",
author = "Henriette Koch and Poul Jennum and Christensen, {Julie Anja Engelhard}",
year = "2018",
doi = "10.1111/jsr.12780",
language = "English",
volume = "27",
journal = "Journal of Sleep Research",
issn = "1365-2869",
publisher = "Wiley-Blackwell",

}

Automatic sleep classification using adaptive segmentation reveals an increased number of rapid eye movement sleep transitions. / Koch, Henriette; Jennum, Poul; Christensen, Julie Anja Engelhard.

In: Journal of Sleep Research, Vol. 27, e12780, 2018.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Automatic sleep classification using adaptive segmentation reveals an increased number of rapid eye movement sleep transitions

AU - Koch, Henriette

AU - Jennum, Poul

AU - Christensen, Julie Anja Engelhard

PY - 2018

Y1 - 2018

N2 - The reference standard for sleep classification uses manual scoring of polysomnography with fixed 30-s epochs. This limits the analysis of sleep pattern, structure and, consequently, detailed association with other physiologic processes. We aimed to improve the details of sleep evaluation by developing a data-driven method that objectively classifies sleep in smaller time intervals. Two adaptive segmentation methods using 3, 10 and 30-s windows were compared. One electroencephalographic (EEG) channel was used to segment into quasi-stationary segments and each segment was classified using a multinomial logistic regression model. Classification features described the power in the clinical frequency bands of three EEG channels and an electrooculographic (EOG) anticorrelation measure for each segment. The models were optimised using 19 healthy control subjects and validated on 18 healthy control subjects. The models obtained overall accuracies of 0.71 ± 0.09, 0.74 ± 0.09 and 0.76 ± 0.08 on the validation data. However, the models allowed a more dynamic sleep, which challenged a true validation against manually scored hypnograms with fixed epochs. The automated classifications indicated an increased number of stage transitions and shorter sleep bouts using models with smaller window size compared with the hypnograms. An increased number of transitions from rapid eye movement (REM) sleep was likewise expressed in the model using 30-s windows, indicating that REM sleep has more fluctuations than captured by today's standard. The models developed are generally applicable and may contribute to concise sleep structure evaluation, research in sleep control and improved understanding of sleep and sleep disorders. The models could also contribute to objective measuring of sleep stability.

AB - The reference standard for sleep classification uses manual scoring of polysomnography with fixed 30-s epochs. This limits the analysis of sleep pattern, structure and, consequently, detailed association with other physiologic processes. We aimed to improve the details of sleep evaluation by developing a data-driven method that objectively classifies sleep in smaller time intervals. Two adaptive segmentation methods using 3, 10 and 30-s windows were compared. One electroencephalographic (EEG) channel was used to segment into quasi-stationary segments and each segment was classified using a multinomial logistic regression model. Classification features described the power in the clinical frequency bands of three EEG channels and an electrooculographic (EOG) anticorrelation measure for each segment. The models were optimised using 19 healthy control subjects and validated on 18 healthy control subjects. The models obtained overall accuracies of 0.71 ± 0.09, 0.74 ± 0.09 and 0.76 ± 0.08 on the validation data. However, the models allowed a more dynamic sleep, which challenged a true validation against manually scored hypnograms with fixed epochs. The automated classifications indicated an increased number of stage transitions and shorter sleep bouts using models with smaller window size compared with the hypnograms. An increased number of transitions from rapid eye movement (REM) sleep was likewise expressed in the model using 30-s windows, indicating that REM sleep has more fluctuations than captured by today's standard. The models developed are generally applicable and may contribute to concise sleep structure evaluation, research in sleep control and improved understanding of sleep and sleep disorders. The models could also contribute to objective measuring of sleep stability.

KW - REM sleep stability

KW - Automatic classification

KW - Data-driven segmentation

KW - Electroencephalography

KW - Sleep stage switching

U2 - 10.1111/jsr.12780

DO - 10.1111/jsr.12780

M3 - Journal article

C2 - 30346084

VL - 27

JO - Journal of Sleep Research

JF - Journal of Sleep Research

SN - 1365-2869

M1 - e12780

ER -