REM sleep behavior disorder (RBD) is characterized by abnormal movements or the physical acting out of dreams during the rapid eye movement (REM) phase of sleep. It is known as ‘isolated’ RBD when it occurs in otherwise healthy adults, and is almost always an early sign of Parkinson’s disease or dementia.
Diagnosing RBD is currently complicated as it can easily go unnoticed or be confused with other conditions. A sleep study known as a video-polysomnogram is required for a definitive diagnosis, which is both costly and challenging to interpret.
Previous studies have suggested that research-grade 3D cameras may be required to detect movements during sleep as blankets may obscure activity. In the current study, however, researchers investigated a machine learning method that uses computer vision to analyze video recordings captured by 2D cameras, which are routinely found in clinical sleep labs, during overnight sleep tests.
For the study, the researchers analyzed 172 overnight video-polysomnogram recordings from a clinical sleep center, including 81 patients with isolated RBD and 91 without, of whom 63 had other sleep disorders such as insomnia and restless leg syndrome, while 28 were healthy sleepers.
The automated machine-learning method ultimately detected isolated RBD with almost 92% accuracy. The algorithm further identified 7 of 11 patients with isolated RBD who did not have noticeable movements during video polysomnograms.
"This automated approach could be integrated into clinical workflow during the interpretation of sleep tests to enhance and facilitate diagnosis, and avoid missed diagnoses. This method could also be used to inform treatment decisions based on the severity of movements displayed during the sleep tests and, ultimately, help doctors personalize care plans for individual patients,” said corresponding author Emmanuel During, MD, Associate Professor of Neurology and Medicine, at the Icahn School of Medicine at Mount Sinai, New York, in a press release.
The researchers wrote in their study that automated analysis of movements should also be tested in home environments. They noted that home monitoring offers a cost advantage compared to in-lab assessments, as well as the possibility of longer monitoring periods, which could improve diagnostic accuracy.
Sources: Science Daily, Annals of Neurology
