Sleep scoring: man vs. machine?

Berthomier, Christian; Brandewinder, Marie

doi:10.1007/s11325-012-0715-1

Cited by 4 publications

(3 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In accordance with recommendations by Berthomier and Brandewinder, 2 statistical measures of agreement, sensitivity, and positive predictive value (PPV) were calculated to assess the performance of Z-PLUS. Cohen’s kappa was used to quantify inter-rater reliability of the visual scoring by pairs of technologists for each study participant, and to evaluate the Z-PLUS versus the PSG Consensus.…”

Section: Methodsmentioning

confidence: 99%

“…Automated scoring of EEG data is a more cost-effective option that removes subjectivity inherent in manual scoring by a technician, 1 but despite greater attention being paid to the development and implementation of technology in recent years, considerable work remains to be done in terms of developing objective, valid, and reliable methods for computing sleep variables. 2 Of the limited research that has been done on automated scoring algorithms, a few demonstrate promising results. A recent study by Malhotra et al 3 evaluated an automated scoring system in which PSG data scored by a computer algorithm was compared against visual scoring by PSG technologists.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Evaluation of an automated single-channel sleep staging algorithm

Wang¹,

Loparo²,

Kelly³

et al. 2015

NSS

View full text Add to dashboard Cite

BackgroundWe previously published the performance evaluation of an automated electroencephalography (EEG)-based single-channel sleep–wake detection algorithm called Z-ALG used by the Zmachine® sleep monitoring system. The objective of this paper is to evaluate the performance of a new algorithm called Z-PLUS, which further differentiates sleep as detected by Z-ALG into Light Sleep, Deep Sleep, and Rapid Eye Movement (REM) Sleep, against laboratory polysomnography (PSG) using a consensus of expert visual scorers.MethodsSingle night, in-lab PSG recordings from 99 subjects (52F/47M, 18–60 years, median age 32.7 years), including both normal sleepers and those reporting a variety of sleep complaints consistent with chronic insomnia, sleep apnea, and restless leg syndrome, as well as those taking selective serotonin reuptake inhibitor/serotonin–norepinephrine reuptake inhibitor antidepressant medications, previously evaluated using Z-ALG were re-examined using Z-PLUS. EEG data collected from electrodes placed at the differential-mastoids (A1–A2) were processed by Z-ALG to determine wake and sleep, then those epochs detected as sleep were further processed by Z-PLUS to differentiate into Light Sleep, Deep Sleep, and REM. EEG data were visually scored by multiple certified polysomnographic technologists according to the Rechtschaffen and Kales criterion, and then combined using a majority-voting rule to create a PSG Consensus score file for each of the 99 subjects. Z-PLUS output was compared to the PSG Consensus score files for both epoch-by-epoch (eg, sensitivity, specificity, and kappa) and sleep stage-related statistics (eg, Latency to Deep Sleep, Latency to REM, Total Deep Sleep, and Total REM).ResultsSensitivities of Z-PLUS compared to the PSG Consensus were 0.84 for Light Sleep, 0.74 for Deep Sleep, and 0.72 for REM. Similarly, positive predictive values were 0.85 for Light Sleep, 0.78 for Deep Sleep, and 0.73 for REM. Overall, kappa agreement of 0.72 is indicative of substantial agreement.ConclusionThis study demonstrates that Z-PLUS can automatically assess sleep stage using a single A1–A2 EEG channel when compared to the sleep stage scoring by a consensus of polysomnographic technologists. Our findings suggest that Z-PLUS may be used in conjunction with Z-ALG for single-channel EEG-based sleep staging.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Evaluation of an automated single-channel sleep staging algorithm

Wang¹,

Loparo²,

Kelly³

et al. 2015

NSS

View full text Add to dashboard Cite

show abstract

“…Fraiwan et al [13] applied a random forest classifier and wavelet features to identify the wakeful state with a 90% accuracy. However, unless applying an extensive number of diverse extracted features, it is difficult to obtain a higher accuracy that is even close to the accuracy levels achieved by experts using manual techniques [16], [17], which is 83 ± 3% [5]. Therefore, automatic sleep classification is still a challenge [17], especially for identifying sleep stages with a single-channel EEG signal.…”

mentioning

confidence: 99%

Analysis and Classification of Sleep Stages Based on Difference Visibility Graphs From a Single-Channel EEG Signal

Zhu

Wen

2014

IEEE J. Biomed. Health Inform.

338

191

View full text Add to dashboard Cite

The existing sleep stages classification methods are mainly based on time or frequency features. This paper classifies the sleep stages based on graph domain features from a single-channel electroencephalogram (EEG) signal. First, each epoch (30 s) EEG signal is mapped into a visibility graph (VG) and a horizontal VG (HVG). Second, a difference VG (DVG) is obtained by subtracting the edges set of the HVG from the edges set of the VG to extract essential degree sequences and to detect the gait-related movement artifact recordings. The mean degrees (MDs) and degree distributions (DDs) P (k) on HVGs and DVGs are analyzed epoch-by-epoch from 14,963 segments of EEG signals. Then, the MDs of each DVG and HVG and seven distinguishable DD values of P (k) from each DVG are extracted. Finally, nine extracted features are forwarded to a support vector machine to classify the sleep stages into two, three, four, five, and six states. The accuracy and kappa coefficients of six-state classification are 87.5% and 0.81, respectively. It was found that the MDs of the VGs on the deep sleep stage are higher than those on the awake and light sleep stages, and the MDs of the HVGs are just the reverse.

show abstract