Sleep state classification using pressure sensor mats.

Area of Science:

• Biomedical Engineering
• Sleep Science
• Sensor Technology

Background:

• Assessing sleep quality and in-bed behavior is crucial for patient monitoring.
• Existing methods may require complex equipment or invasive sensors.
• There is a need for non-invasive, accurate sleep state detection systems.

Purpose of the Study:

• To develop a novel sleep state classification approach using only surface pressure sensors.
• To introduce a new 'pre-wake' state classification for more granular sleep monitoring.
• To evaluate the accuracy of the proposed method for identifying sleep, pre-wake, and wake states.

Main Methods:

• A mobility metric was defined using successive surface pressure sensor body maps.
• Statistical features were computed based on the derived mobility metric.
• A customized random forest classifier was employed for sleep state identification.

Main Results:

• The algorithm achieved 96.1% accuracy for two-class identification (sleep, wake).
• The algorithm achieved 88% accuracy for three-class identification (sleep, pre-wake, wake).
• The proposed method demonstrates high efficacy in distinguishing sleep states.

Conclusions:

• Surface pressure sensors can be effectively utilized for non-invasive sleep state detection.
• The novel approach accurately classifies sleep, pre-wake, and wake states.
• This method offers a promising tool for enhanced sleep quality assessment and behavior monitoring.