Unobtrusive sleep posture estimation using pressure sensor in home sleep.

Area of Science:

• Biomedical Engineering
• Sleep Science
• Wearable Technology

Background:

• Sleep posture significantly impacts sleep health and quality, influencing physiological indicators.
• Existing posture estimation methods often lack validation in real-world settings.
• Accurate, noninvasive sleep monitoring is crucial for understanding sleep disorders and improving patient outcomes.

Purpose of the Study:

• To develop and validate a novel method for determining sleep posture in real-world environments using pressure sensor data.
• To assess the feasibility of implementing sleep monitoring technologies in daily life and clinical contexts.
• To contribute to the development of explainable, noninvasive, long-term sleep monitoring systems.

Main Methods:

• A support vector machine (SVM) algorithm was trained using force-sensitive resistor (FSR) data from 22 participants in a laboratory setting.
• The SVM classified four sleep postures (supine, left-lateral, right-lateral, prone) based on extracted features (area, curvature, row length ratio).
• The algorithm's performance was evaluated using FSR data from 10 participants in their home environments.

Main Results:

• The SVM model achieved 78.1% accuracy and a Cohen's kappa of 0.71 on laboratory data.
• When applied to home-environment data, the method demonstrated higher performance with 86.1% accuracy and a Cohen's kappa of 0.76.
• The model successfully classified four distinct sleep postures in uncontrolled, real-world conditions.

Conclusions:

• The sleep posture estimation model trained in a laboratory setting maintains high performance in real-world conditions.
• This validates the feasibility of using pressure sensor-based technology for noninvasive, long-term sleep monitoring.
• The findings support the potential for clinical applications in embedded systems and hospital environments, emphasizing explainable feature-based models.