Non-invasive sensors based human state in nightlong sleep analysis for home-care

In this paper we present methodology used in a non-invasive, easy-to-use and low-cost monitoring system for nightlong human sleep quantification. Our system uses simultaneous measurement of three different signals representing the activity of the human body: infrared video-recorded subject motion, audio-recorded acoustic effects and the three-leads electrocardiogram. Signal-specific interpretation methods yield parameters selected as most discriminative for the sleep quality, synchronized and combined as a sleep record. In the experimental stage the nightlong sleep was supervised by the reference EEG recordings and particular components of the sleep record were correlated to the presence of delta wave representing deep sleep. Significant correlation values in most subjects allow to validate the proposed sleep record as comparable to the standard polysomnogram.