Overnight Sleep Staging Using Chest-Worn Accelerometry

Area of Science:

• Sleep Medicine
• Biomedical Engineering
• Signal Processing

Background:

• Overnight sleep staging is crucial for diagnosing sleep disorders.
• Polysomnography (PSG) is the gold standard but is obtrusive.
• Less-obtrusive sensing modalities are emerging for sleep analysis.

Purpose of the Study:

• To develop and validate an algorithm for "proxy" sleep staging.
• To utilize cardiac and respiratory signals from a chest-worn accelerometer.
• To assess the feasibility of unobtrusive sleep staging.

Main Methods:

• Collected data from 323 participants across two sleep centers using PSG and a chest-worn accelerometer.
• Derived cardiac and respiratory features from accelerometer data.
• Applied an automatic cardio-respiratory sleep staging method and compared results to PSG.

Main Results:

• Achieved 80.8% accuracy and Cohen's kappa of 0.68 for four-class sleep staging (Wake, REM, N1+N2, N3).
• Reached 93.3% accuracy for Wake vs. Sleep classification, with 78.7% sensitivity and 96.6% specificity.
• Demonstrated reliable sleep staging across diverse age, BMI, and sleep disorder populations.

Conclusions:

• Cardiorespiratory signals from chest-worn accelerometers can effectively estimate sleep stages.
• This approach offers a less-obtrusive alternative to traditional PSG for sleep staging.
• The findings support clinical applications in sleep medicine, enabling easier sleep disorder diagnosis.