Automatic evaluation of polysomnographic recordings in Matlab

The article describes the algorithm created in the interactive environment Matlab and provides the detection of two sleep stages, specifically NREM1 and NREM3/4. One of possible methods is to split the signal into 30 seconds segments. These are later divided by means of adaptive segmentation. Subsequently the properties of individual segments are described (using Hjorth parameters and amplitude) and on the basis of these properties using cluster analysis the segments are classified into the classes. Consequent processing and evaluation leads to determination of the sleep states, of which the hypnogram is created at the end.