An electroencephalography-based sleep index and supervised machine learning as a suitable tool for automated sleep
Eris van Twist1, Floor W Hiemstra2,3, Arnout B G Cramer1
1Department of Neonatal and Pediatric Intensive Care, Division of Pediatric Intensive Care, Erasmus MC Sophia Children's Hospital, Rotterdam, The Netherlands.
Insights
A new electroencephalography (EEG)-based index enables automated sleep monitoring in children. This method accurately classifies sleep states at the bedside, improving pediatric sleep research.
Area of Science:
- Pediatric intensive care research
- Sleep medicine
- Biomedical engineering
Background:
- Sleep disruption is common in pediatric intensive care units (PICUs).
- Current bedside sleep monitoring methods are limited.
- Electroencephalography (EEG) offers potential for real-time sleep analysis.
Purpose of the Study:
- To develop and validate a novel EEG-based index for automated sleep classification in children.
- To assess the index's performance across different age groups and EEG channels.
- To enable real-time, non-invasive sleep monitoring at the pediatric bedside.
Main Methods:
- Retrospective analysis of polysomnography recordings from non-critically ill children (6 months to 18 years).
- Calculation of spectral band powers from EEG data to derive a sleep index (gamma to delta power ratio).
- Development of sleep classification models (2, 3, and 4 states) using decision trees and cross-validation.
Main Results:
- The gamma to delta spectral power ratio using F4-A1 and F3-A1 EEG channels demonstrated the best performance.
- Balanced accuracy reached 0.88 for two-state and 0.74 for three-state sleep classification.
- High performance was consistent across different age categories (6 months to 18 years).
Conclusions:
- An interpretable and generalizable EEG-based sleep index was developed for automated bedside monitoring.
- The index shows good performance for two- and three-state sleep classification in children.
- This approach facilitates continuous, non-invasive sleep assessment in pediatric settings.
Study Objectives:
Although sleep is frequently disrupted in the pediatric intensive care unit, it is currently not possible to perform real-time sleep monitoring at the bedside. In this study, spectral band powers of electroencephalography data are used to derive a simple index for sleep classification.
Methods:
Retrospective study at Erasmus MC Sophia Children's Hospital, using hospital-based polysomnography recordings obtained in non-critically ill children between 2017 and 2021. Six age categories were defined: 6-12 months, 1-3 years, 3-5 years, 5-9 years, 9-13 years, and 13-18 years. Candidate index measures were derived by calculating spectral band powers in different frequent frequency bands of smoothed electroencephalography. With the best performing index, sleep classification models were developed for two, three, and four states via decision tree and five-fold nested cross-validation. Model performance was assessed across age categories and electroencephalography channels.
Results:
In total 90 patients with polysomnography were included, with a mean (standard deviation) recording length of 10.3 (1.1) hours. The best performance was obtained with the gamma to delta spectral power ratio of the F4-A1 and F3-A1 channels with smoothing. Balanced accuracy was 0.88, 0.74, and 0.57 for two-, three-, and four-state classification. Across age categories, balanced accuracy ranged between 0.83 and 0.92 and 0.72 and 0.77 for two- and three-state classification, respectively.
Conclusions:
We propose an interpretable and generalizable sleep index derived from single-channel electroencephalography for automated sleep monitoring at the bedside in non-critically ill children ages 6 months to 18 years, with good performance for two- and three-state classification.
Citation:
van Twist E, Hiemstra FW, Cramer ABG, et al. An electroencephalography-based sleep index and supervised machine learning as a suitable tool for automated sleep classification in children. J Clin Sleep Med. 2024;20(3):389-397.
