Detection of obstructive sleep apnea from wearable physiological devices

Abstract

AbstractObstructive sleep apnea (OSA) is a common respiratory condition characterized by respiratory tract obstruction and breathing disorder. Early detection and treatment of OSA can significantly reduce morbidity and mortality. OSA is often diagnosed with overnight polysomnography (PSG) monitoring; however, continuous PSG monitoring is unfeasible as it is costly and uncomfortable for patients. To circumvent these issues, we propose a detection method of OSA events, named DRIVEN, using only two signals that can be easily measured at home: abdominal movement and pulse oximetry. On test data, DRIVEN achieves an accuracy and F1-score of 88%, a reasonable trade-off between the model ‘s performance and patient ‘s comfort. We use data from three sleep studies from the National Sleep Research Resource (NSRR), the largest public repository, consisting of 10,878 recordings. DRIVEN is based on a combination of deep convolutional neural networks and a light-gradient-boost machine for classification. Since DRIVEN is simple and computationally efficient, we expect that it can be implemented for automatic detection of OSA in unsupervised long-term home monitoring systems, reducing costs to healthcare systems and improving patient care.