Dataset Information

Quasi-supervised scoring of human sleep in polysomnograms using augmented input variables.

ABSTRACT: The limitations of manual sleep scoring make computerized methods highly desirable. Scoring errors can arise from human rater uncertainty or inter-rater variability. Sleep scoring algorithms either come as supervised classifiers that need scored samples of each state to be trained, or as unsupervised classifiers that use heuristics or structural clues in unscored data to define states. We propose a quasi-supervised classifier that models observations in an unsupervised manner but mimics a human rater wherever training scores are available. EEG, EMG, and EOG features were extracted in 30s epochs from human-scored polysomnograms recorded from 42 healthy human subjects (18-79 years) and archived in an anonymized, publicly accessible database. Hypnograms were modified so that: 1. Some states are scored but not others; 2. Samples of all states are scored but not for transitional epochs; and 3. Two raters with 67% agreement are simulated. A framework for quasi-supervised classification was devised in which unsupervised statistical models-specifically Gaussian mixtures and hidden Markov models--are estimated from unlabeled training data, but the training samples are augmented with variables whose values depend on available scores. Classifiers were fitted to signal features incorporating partial scores, and used to predict scores for complete recordings. Performance was assessed using Cohen's ? statistic. The quasi-supervised classifier performed significantly better than an unsupervised model and sometimes as well as a completely supervised model despite receiving only partial scores. The quasi-supervised algorithm addresses the need for classifiers that mimic scoring patterns of human raters while compensating for their limitations.

SUBMITTER: Yaghouby F

PROVIDER: S-EPMC4447106 | biostudies-literature | 2015 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Quasi-supervised scoring of human sleep in polysomnograms using augmented input variables.

Yaghouby Farid F Sunderam Sridhar S

Computers in biology and medicine 20150123

The limitations of manual sleep scoring make computerized methods highly desirable. Scoring errors can arise from human rater uncertainty or inter-rater variability. Sleep scoring algorithms either come as supervised classifiers that need scored samples of each state to be trained, or as unsupervised classifiers that use heuristics or structural clues in unscored data to define states. We propose a quasi-supervised classifier that models observations in an unsupervised manner but mimics a human ...[more]

PMID: 25679475

Similar Datasets

Project description:BackgroundDifferentiating nonwear time from sleep and wake times is essential for the estimation of sleep duration based on actigraphy data. To efficiently analyze large-scale data sets, an automatic method of identifying these three different states is required. Therefore, we developed a classification algorithm to determine nonwear, sleep and wake periods from accelerometer data. Our work aimed to (I) develop a new pattern recognition algorithm for identifying nonwear periods from actigraphy data based on the influence of respiration rate on the power spectrum of the acceleration signal and implement it in an automatic classification algorithm for nonwear/sleep/wake states; (II) address motion artifacts that occur during nonwear periods and are known to cause misclassification of these periods; (III) adjust the algorithm depending on the sensor position (wrist, chest); and (IV) validate the algorithm on both healthy individuals and patients with sleep disorders.MethodsThe study involved 98 participants who wore wrist and chest acceleration sensors for one day of measurements. They spent one night in the sleep laboratory and continued to wear the sensors outside of the laboratory for the remainder of the day. The results of the classification algorithm were compared to those of the reference source: polysomnography for wake/sleep and manual annotations for nonwear/wear classification.ResultsThe median kappa values for the two locations were 0.83 (wrist) and 0.84 (chest). The level of agreement did not vary significantly by sleep health (good sleepers vs. subjects with sleep disorders) (p = 0.348, p = 0.118) or by sex (p = 0.442, p = 0.456). The intraclass correlation coefficients of nonwear total time between the reference and the algorithm were 0.92 and 0.97 with the outliers and 0.95 and 0.98 after the outliers were removed for the wrist and chest, respectively. There was no evidence of an association between the mean difference (and 95% limits of agreement) and the mean of the two methods for either sensor position (wrist p = 0.110, chest p = 0.164), and the mean differences (algorithm minus reference) were 5.11 [95% LoA -15.4-25.7] and 1.32 [95% LoA -9.59-12.24] min/day, respectively, after the outliers were removed.DiscussionWe studied the influence of the respiration wave on the power spectrum of the acceleration signal for the differentiation of nonwear periods from sleep and wake periods. The algorithm combined both spectral analysis of the acceleration signal and rescoring. Based on the Bland-Altman analysis, the chest-worn accelerometer showed better results than the wrist-worn accelerometer.

Project description:Background: Performance of wrist actigraphy in assessing sleep not only depends on the sensor technology of the actigraph hardware but also on the attributes of the interpretative algorithm (IA). The objective of our research was to improve assessment of sleep quality, relative to existing IAs, through development of a novel IA using deep learning methods, utilizing as input activity count and heart rate variability (HRV) metrics of different window length (number of epochs of data). Methods: Simultaneously recorded polysomnography (PSG) and wrist actigraphy data of 222 participants were utilized. Classic deep learning models were applied to: (a) activity count alone (without HRV), (b) activity count + HRV (30-s window), (c) activity count + HRV (3-min window), and (d) activity count + HRV (5-min window) to ascertain the best set of inputs. A novel deep learning model (Haghayegh Algorithm, HA), founded on best set of inputs, was developed, and its sleep scoring performance was then compared with the most popular University of California San Diego (UCSD) and Actiwatch proprietary IAs. Results: Activity count combined with HRV metrics calculated per 5-min window produced highest agreement with PSG. HA showed 84.5% accuracy (5.3-6.2% higher than comparator IAs), 89.5% sensitivity (6.2% higher than UCSD IA and 6% lower than Actiwatch proprietary IA), 70.0% specificity (8.2-34.3% higher than comparator IAs), and 58.7% Kappa agreement (16-23% higher than comparator IAs) in detecting sleep epochs. HA did not differ significantly from PSG in deriving sleep parameters-sleep efficiency, total sleep time, sleep onset latency, and wake after sleep onset; moreover, bias and mean absolute error of the HA model in estimating them was less than the comparator IAs. HA showed, respectively, 40.9% and 54.0% Kappa agreement with PSG in detecting rapid and non-rapid eye movement (REM and NREM) epochs. Conclusions: The HA model simultaneously incorporating activity count and HRV metrics calculated per 5-min window demonstrates significantly better sleep scoring performance than existing popular IAs.

Dataset Information

Quasi-supervised scoring of human sleep in polysomnograms using augmented input variables.

Publications

Quasi-supervised scoring of human sleep in polysomnograms using augmented input variables.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets