A computerized method comprises statistically analyzing, using one or more processors, at least one section of a digitally recorded electroencephalography (EEG) signal that comprises an arousal segment to determine, for the arousal segment, at least one of amplitude and power at different frequencies as a function of time； selecting specified features from the results of the analysis and normalizing the selected features； and assigning, using one or more processors, an intensity scale value to the arousal segment based on the normalized selected features and a reference data set, the reference data set comprising normalized features corresponding to the normalized selected features and being generated based on a plurality of EEG signals comprising arousal segments to which intensity scale values have been assigned based on a visual inspection of the EEG signals.
