ABSTRACT
Sleep state scoring usually relies on polysomnographic measurements, which include electroencephalogram (EEG), electromyogram (EMG), electro-oculogram (EOG), two or three lead chest electrocardiogram (ECG), and may include other measurements. Overall, polysomnography is an intrusive procedure not well tolerated by infants and elderly. The goal of this research is to study possibility of automatic sleep state scoring from less intrusive measurements such as activity measurements and respiratory measurements by inductive plethysmography. The study is based on the Collaborative Home Infant Monitoring Evaluation (CHIME) dataset. Results demonstrate that the suggested approach is capable of scoring sleep states (awake, rapid eye movement and quiet sleep) with good accuracy.
