Nightcap: A Home-Based Sleep Monitoring System

This is the authors' abstract. We don't add key points for this paper.

In an attempt to offer a home-based adjunct to traditional sleep laboratory methods, we developed a system to monitor sleep, and to predict algorithmically non-rapid-eye-movement (NREM) and rapid-eye-movement (REM) sleep states, using eye and body motility as the only parameters. Eye movement was measured using a strain gauge transducer applied to the eyelid of subjects, while body movement was measured using a piezo-ceramic phono cartridge. Both transducers were mounted on a tennis headband, along with electronics that amplified, filtered, and digitized the signals. Digital pulse signals were input to a portable computer in minute-long epochs, and state-predicting algorithms were run based on this motility data. Four subjects were monitored in the sleep lab with both our headgear and standard polysomnography. Hand-scored sleep records were compared with those predicted by computer algorithms. Algorithm-predicted states agreed with hand-scored ones an average of 85.57% (SEM +/- 1.7%). Mean values for sleep onset and REM latency were within 1.6 and 10.8 min of polysomnographic records, respectively. These results are encouraging, and suggest that this system could provide a comfortable, subject operable, and inexpensive method for the evaluation of sleep at home.