A Novel Deep Neural Network for Sleep Apnea Event Detection with Pressure Signals

Accurately detecting obstructive sleep apnea (OSA) is critical for patients to assess their sleep conditions. Pressure signals from a sensor array embedded in a mat is a promising alternative for sleep monitoring in a contactless way. In this study, we propose a novel OSA detection method called MaxEng-SleepN et by utilizing the maximum energy for integrating the multiple channel pressure signals. Meanwhile, we perform a pilot study to recruit volunteers from the Peking Union Medical College Hospital to obtain overnight pressure signals. Experiment results demonstrate that our proposed MaxEng-SleepN et can achieve a competitive result. It means that the MaxEng-SleepNet model as a pivotal component has the potential to be embedded into a healthcare monitoring system to offer a sleep monitoring service.