A novel approach to automatic sleep stage classification using forehead electrophysiological signals
Hengyan Guo1, Yang Di1, Xingwei An1,2
1Academy of Medical Engineering and Translational Medicine, Tianjin University, Tianjin, China.
Heliyon
|January 2, 2023
Summary
This study introduces a portable automatic sleep scoring system using forehead electrophysiological signals, achieving 90.25% accuracy. The novel method enhances sleep disorder diagnosis and monitoring capabilities.
Area of Science:
- Biomedical Engineering
- Neuroscience
- Sleep Medicine
Background:
- Automatic sleep stage scoring is crucial for diagnosing and managing sleep disorders.
- Current methods suffer from poor model generalization and require non-portable equipment.
Purpose of the Study:
- To develop a novel, effective, and convenient automatic sleep scoring system using forehead electrophysiological signals.
- To improve the portability and generalizability of automatic sleep staging.
Main Methods:
- Utilized three forehead electrophysiological signals: Fh1, Fh2 (electroencephalogram), and Fhz (electrooculogram).
- Extracted spectral, statistical, and entropy features using discrete wavelet transform (DWT).
- Employed Light Gradient Boosting Machine (LGB), Random Forest (RF), and Support Vector Machine (SVM) for classifying four sleep stages (awake, LS, DS, REM).
Main Results:
- The system achieved an overall classification accuracy of 90.25% with a kappa coefficient of 0.857.
- Performance was validated on Sleep-EDFX and Own-data databases, including polysomnograms (PSGs) and forehead signals from 28 subjects.
- The combination of Fh1, Fh2, and Fhz signals yielded state-of-the-art results.
Conclusions:
- The proposed forehead-based system offers high portability and effective multichannel sleep stage scoring.
- This advancement facilitates future long-term sleep quality monitoring applications.
