Comparison of computerized methods for rapid eye movement sleep without atonia detection

Area of Science:

• Neurology
• Sleep Medicine
• Biomedical Engineering

Background:

• Diagnosis of REM sleep behavior disorder (RBD) relies on detecting REM sleep without atonia.
• Current gold standard visual scoring is time-consuming and subjective.
• Automated methods offer potential for more efficient and objective assessment.

Purpose of the Study:

• To compare the diagnostic performance of various automated indices for REM sleep without atonia detection.
• To evaluate the influence of arousal and apnea events on automated REM sleep without atonia detection algorithms.
• To identify the most effective automated method for diagnosing RBD.

Main Methods:

• Five automated algorithms (RAI, supra-threshold-REM-activity, Frandsen, short/long muscle activity, Kempfner) were applied to polysomnography data.
• Indices were calculated across six configurations, excluding or including specific muscle activities (arousals, apnea events).
• Discrimination capabilities were tested across multiple patient groups (Controls, PD-no RBD, PD+RBD, iRBD, PLMD) and comparisons.

Main Results:

• No single algorithm demonstrated optimal performance across all configurations and comparisons.
• The REM atonia index (RAI) showed the most consistent and sensible results for RBD detection.
• Apnea and arousal-related movements did not significantly affect the classification accuracy of the algorithms.

Conclusions:

• Current automated methods for REM sleep without atonia detection require further refinement.
• RAI is a promising index for RBD detection, warranting further investigation.
• Automated detection of REM sleep without atonia can be robust to common confounding events like arousals and apneas.