Visualization and clustering of sleep states in a frequency domain feature space

Sleep studies assess the recurrent manifestation of stereotype configurations of relevant biosignals. These configurations are known as states (Wake, REM sleep and NonREM sleep) and stages (N1-N3 within NREM sleep). These two fundamental descriptive domains, time course and variable configuration, can be readily rendered available through improved visualization techniques. Time course is summarized by EEG spectrograms, instantaneous frequency analysis of cardio-respiratory signals and other sleep dependent variables. State and stage configurations can be further evidenced as clusters in 2D or 3D spaces whose axis are sleep-relevant extracted variables. The latter techniques also allows for visualization of transition process as pathways from one cluster to another.