A sleep stage estimation algorithm based on cardiorespiratory signals derived from a suprasternal pressure sensor.

Area of Science:

• Biomedical Engineering
• Sleep Medicine
• Signal Processing

Background:

• Portable sleep monitoring systems are crucial for home-based assessments.
• Maintaining accuracy of physiological signals like respiration is vital during this transition.
• Suprasternal pressure (SSP) sensors, used for sleep-disordered breathing (SDB), can also capture cardiac vibrations.

Purpose of the Study:

• To evaluate the use of cardiorespiratory signals from an SSP sensor for automatic sleep stage estimation.
• To determine if separating respiratory and cardiac signals enhances sleep staging accuracy.
• To assess the potential of SSP sensor data for advanced clinical sleep disorder assessments.

Main Methods:

• Collected SSP sensor signals from 100 adults undergoing polysomnography.
• Separated respiratory effort and cardiac activity signals from the mixed SSP data.
• Utilized a neural network trained on these signals for sleep stage estimation.
• Compared results with manual sleep scoring, including Cohen's kappa and total sleep time.

Main Results:

• The mixed SSP signal showed moderate agreement (kappa 0.53-0.62) with manual scoring.
• Decoupling signals and using cardiac data for heart rate estimation improved agreement (kappa 0.63-0.71).
• The method demonstrated high accuracy, specificity, and sensitivity for sleep staging.
• Total sleep time estimation had a small average error (-1.83 min) but a wide confidence interval (±55 min).

Conclusions:

• SSP sensor cardiorespiratory signals can be effectively used for automatic sleep stage estimation.
• Separating respiratory and cardiac components significantly enhances the accuracy of portable sleep monitoring.
• This approach offers potential for developing compact, information-rich systems for clinical sleep disorder assessment and monitoring.