Sleep Apnea Detection Using Oxygen Saturation During Nocturnal Sleep

The early detection and treatment of sleep apnea syndrome (SAS) is becoming an important issue in medical practice, as the conventional polysomnography-based method is not only costly, but also demands high expertise from technicians. In this study, we propose a method for diagnosing SAS using noninvasive measurements of oxygen saturation (SpO2), obtained from a pulse oximeter as the input signal, using convolutional neural networks (CNNs). As a result of 10-fold cross-validation, we confirmed that the proposed method yields an 88.99% accuracy.