Sleep Staging Using Deep Neural Networks

Abstract

Polysomnography (PSG) is the current state of the art for sleep staging but has several drawbacks: it requires a multitude of sensors and needs experienced technicians, which makes it expensive. It is further not suited for long-term monitoring in ambulatory environments [1]. Recent advances in optical heart rate monitoring allow for devices which are low-cost and suitable for ambulatory sleep staging. Even though not expected to completely replace PSG, such devices would allow for reliable pre-screening of several consecutive nights of larger populations.