Knowledge-based identification of sleep stages based on two forehead electroencephalogram channels

摘要: Sleep quality is important, especially given the considerable number of sleep-related pathologies. The distribution sleep stages a highly effective and objective way quantifying quality. As standard multi-channel recording used in study sleep, polysomnography (PSG) widely diagnostic scheme medicine. However, process clinical test, including PSG manual scoring, complex, uncomfortable, time-consuming. This difficult to implement when taking whole measurements at home for general healthcare purposes. work presents novel stage classification system, based on features from two forehead EEG channels FP1 FP2. By forehead, where there no hair, proposed system can monitor physiological changes during more practical than previous systems. Through headband or self-adhesive technology, necessary sensors be applied easily by users home. Analysis results demonstrate that performance overcomes individual differences between different participants terms automatically classifying stages. Additionally, identify kernel extracted EEG, which are closely related with clinician’s expert knowledge. Moreover, classified into five using relevance vector machine. In leave-one-subject-out cross validation analysis, we found our correctly classify an average accuracy 76.7 ± 4.0 (SD) % (average kappa 0.68 0.06 (SD)). Importantly, viable alternative measuring signals conveniently evaluate reliably, ultimately improving public healthcare.