A Novel Tool for Sequential Fusion of Nonlinear Features: A Sleep Psychology Application

摘要

A framework for automated scoring of sleep stages during afternoon naps of healthy humans is introduced. This is achieved by sequential fusion of nonlinear features extracted from three physiological channels: the electroencephalogram (EEG), electrooculogram (EOG) and respiratory trace (RES). These features are generated by means of the recently introduced "Delay Vector Variance" (DVV) method which examines local predictability of a signal in phase space. The analysis is accompanied by a set of comprehensive simulations, supporting the approach.