Sleep staging based on signals acquired through bed sensor.

Area of Science:

• Biomedical Engineering
• Sleep Science
• Signal Processing

Background:

• Accurate sleep staging is crucial for diagnosing sleep disorders.
• Traditional polysomnography is expensive and complex.
• Non-invasive, cost-effective sleep monitoring solutions are needed.

Purpose of the Study:

• To develop and validate a novel system for sleep macrostructure evaluation using bed sensor data.
• To assess the accuracy of a system combining heart-beat interval (HBI) and movement activity for sleep staging.

Main Methods:

• Utilized Emfit sensor foils integrated into a mattress to capture heart-beat interval (HBI) and movement activity.
• Employed a time-variant autoregressive model (TVAM) for feature extraction from sensor data.
• Implemented a hidden Markov model (HMM) for sleep stage classification (wake, NREM, REM).

Main Results:

• The system achieved a total accuracy of approximately 79% in classifying sleep stages against expert polysomnography.
• Kappa index values ranged from 0.43 to 0.44, indicating moderate agreement.
• The system demonstrated reliable performance using a limited set of HBI and movement features.

Conclusions:

• The combination of HBI and movement features derived from bed sensors offers a viable and cost-effective approach for sleep staging.
• This system presents a simpler alternative to conventional polysomnography for evaluating sleep macrostructure.
• Further validation may establish this method as a practical tool in sleep research and clinical settings.