Selection of Measures for Sleep Stages Classification
K. Šušmáková, Anna Krakovská · 2009
In this work a large amount of features from polysomnographic recordings was tested to find the best set of variables for sleep stages classification. Discriminant analysis was done with Fisher quadratic classifier and forward selection procedure. Resulting set contains 14 measures from EEG, EOG, EMG and ECG signals, some of them are used in this context for the first time (e. g. fractal exponent and entropy of EMG).