Datasets for Graph Analysis of Single-Channel Sleep Stage EEG with Different Electrode Placements

Complex networks have been successfully applied to sleep stage analysis and classification. However, whether the electroencephalogram (EEG) montage reference will affect the network properties is still unclear. This study investigates network changes and evaluates the sleep stage classification performance using independent subjects and non-overlap epochs two types of testing sets respectively, while the raw EEG signals were recorded from five individual channels sleep EEG signals using bipolar and monopolar montages, respectively. Three types of network features: average degree(D), mean clustering coefficient (C), and shortest path length (L) are extracted.