Assessing REM Sleep Behaviour Disorder: From Machine Learning Classification to the Definition of a Continuous Dissociation Index.

Objectives: Rapid Eye Movement Sleep Behaviour Disorder (RBD) is regarded as a prodrome of neurodegeneration, with a high conversion rate to α-synucleinopathies such as Parkinson's Disease (PD). The clinical diagnosis of RBD co-exists with evidence of REM Sleep Without Atonia (RSWA), a parasomnia that features loss of physiological muscular atonia during REM sleep. The objectives of this study are to implement an automatic detection of RSWA from polysomnographic traces, and to propose a continuous index (the Dissociation Index) to assess the level of dissociation between REM sleep stage and atonia. This is performed using Euclidean distance in proper vector spaces. Each subject is assigned a dissociation degree based on their distance from a reference, encompassing healthy subjects and clinically diagnosed RBD patients at the two extremes. Methods: Machine Learning models were employed to perform automatic identification of patients with RSWA through clinical polysomnographic scores, together with variables derived from electromyography. Proper distance metrics are proposed and tested to achieve a dissociation measure. Results: The method proved efficient in classifying RSWA vs. not-RSWA subjects, achieving an overall accuracy, sensitivity and precision of 87%, 93% and 87.5%, respectively. On its part, the Dissociation Index proved to be promising in measuring the impairment level of patients. Conclusions: The proposed method moves a step forward in the direction of automatically identifying REM sleep disorders and evaluating the impairment degree. We believe that this index may be correlated with the patients' neurodegeneration process; this assumption will undergo a robust clinical validation process involving healthy, RSWA, RBD and PD subjects.

Single-channel EEG classification of sleep stages based on REM microstructure.

Rapid-eye movement (REM) sleep, or paradoxical sleep, accounts for 20-25% of total night-time sleep in healthy adults and may be related, in pathological cases, to parasomnias. A large percentage of Parkinson's disease patients suffer from sleep disorders, including REM sleep behaviour disorder and hypokinesia; monitoring their sleep cycle and related activities would help to improve their quality of life. There is a need to accurately classify REM and the other stages of sleep in order to properly identify and monitor parasomnias. This study proposes a method for the identification of REM sleep from raw single-channel electroencephalogram data, employing novel features based on REM microstructures. Sleep stage classification was performed by means of random forest (RF) classifier, K-nearest neighbour (K-NN) classifier and random Under sampling boosted trees (RUSBoost); the classifiers were trained using a set of published and novel features. REM detection accuracy ranges from 89% to 92.7%, and the classifiers achieved a F-1 score (REM class) of about 0.83 (RF), 0.80 (K-NN), and 0.70 (RUSBoost). These methods provide encouraging outcomes in automatic sleep scoring and REM detection based on raw single-channel electroencephalogram, assessing the feasibility of a home sleep monitoring device with fewer channels.