LAUSR

Objective. Sleep monitoring by polysomnography (PSG) severely degrades sleep quality. In order to reduce the load of sleep monitoring, an approach to automatic sleep stage classification without an electroencephalogram (EEG) was proposed. Approach. A total of 124 records from the public dataset ISRUC-Sleep incorporating American Academy of Sleep Medicine (AASM) standards were used: 10 records were from the healthy group while the others were from sleep disorder groups. The 124 records were collected from 116 subjects (eight subjects had two records each, the others had one record each) with ages ranging from 20 to 85 years. A total of 108 features were extracted from the two-channel electrooculograms (EOGs) and six features were extracted from the one-channel electromyogram (EMG). A novel ‘quasi-normalization’ method was proposed and used for feature normalization. Then the random forest algorithm was used to classify five stages, including wakefulness, rapid eye movement sleep, N1 sleep, N2 sleep and N3 sleep. Main results. Using 114 normalized features from the combination of EOG (108 features) and EMG (6 features) data, Cohen’s kappa coefficient was 0.749 and the accuracy was 80.8% by leave-one-out cross-validation. As a reference for AASM standards using a computer-assisted method, Cohen’s kappa coefficient was 0.801 and the accuracy was 84.7% for the same dataset based on 438 normalized features from a combination of EEG (324 features), EOG (108 features) and EMG (6 features) data. Significance. A combination of EOG and EMG can reduce the load of sleep monitoring, and achieves comparable performance to the ‘gold standard’ signals of EEG, EOG and EMG for sleep stage classification.