Validation of RBDtector: An Open-Source Automated Software for Scoring REM Sleep Without Atonia.

Abstract

Accurate quantification of REM sleep without atonia (RSWA) is essential in the diagnosis of idiopathic/isolated REM sleep behaviour disorder (iRBD). This study aims to validate RBDtector, a free and open-source tool for automated RSWA quantification using the Sleep Innsbruck Barcelona (SINBAR) scoring method, by comparing its performance against human visual scoring in a large independent cohort of subjects with iRBD and healthy controls. Muscle activity from 118 iRBD participants and 37 healthy controls that underwent polysomnography was analysed by RBDtector and compared with human visual scoring. Diagnostic performance was evaluated using receiver operating characteristic curves, and optimal cut-offs for iRBD screening and diagnosis were determined. The results of RSWA quantification were applied to survival analyses of time to phenoconversion. RBDtector showed excellent agreement with human visual scoring, particularly in 'any' RSWA activity (Pearson's correlation = 0.89, R² = 0.79, p < 0.001). RBDtector identified iRBD subjects with 95.6% sensitivity and 95.5% specificity by using a cutoff of 33.0% for combined 'any' RSWA activity in the submentalis and flexor digitorum superficialis muscles, with each muscle in isolation providing comparable results. In iRBD patients, each 10% increase in submentalis 'any' activity was associated with a 23% increase in the risk of phenoconversion (HR = 1.23, 95% CI [1.06, 1.44], p = 0.008), while no associations were observed with increased activity in the flexor digitorum superficialis or tibialis anterior. RBDtector provides accurate, automated RSWA quantification comparable to human visual scoring, offering a reliable and efficient method to support the diagnosis of iRBD and identify iRBD at a higher risk of phenoconversion.