Detecting REM sleep from the finger: an automatic REM sleep algorithm based on peripheral arterial tone (PAT) and...

Area of Science:

• Biomedical Engineering
• Sleep Medicine
• Signal Processing

Background:

• Polysomnography (PSG) for REM sleep scoring is labor-intensive.
• Ambulatory sleep monitoring requires automated, cost-effective solutions.
• Traditional REM detection relies on EEG, EOG, and EMG.

Purpose of the Study:

• Develop an automatic REM sleep detection algorithm using peripheral arterial tone (PAT) and actigraphy.
• Validate the algorithm's performance against PSG gold standard.
• Enable reliable unattended ambulatory sleep monitoring.

Main Methods:

• Algorithm developed using PAT and actigraphy from Watch-PAT100 device.
• Features extracted from PAT amplitude and inter-pulse periods over 5-min intervals.
• Genetic algorithm (GA) optimized prediction function for REM epoch likelihood.
• Trained and validated on separate sets of 30 patients each.

Main Results:

• The algorithm achieved 78% sensitivity, 92% specificity, and 89% agreement.
• Demonstrated robust performance on a validation set.
• Genetic algorithm optimization proved effective.

Conclusions:

• The developed algorithm offers a reliable, automated method for REM sleep detection.
• PAT and actigraphy provide a viable alternative to traditional PSG for REM scoring.
• Potential for widespread use in ambulatory and home-based sleep studies.