Exacerbation in Obstructive Sleep Apnea: Early Detection and Monitoring Using a Single Channel EEG with Quadratic Discriminant Analysis

Abstract

Exacerbation monitoring of obstructive sleep apnea (OSA) is important for the evaluation of treatment effectiveness and tracking the disease progression. In this study, we investigated the use of spectral features from single channel electroencephalography (EEG) for early detection and monitoring of OSA exacerbation using the Sleep Health Heart Study dataset. We have explored 22 features at different sleep stages corresponding to different frequency bands to distinguish 410 subjects in the stable and exacerbation groups. An optimal set of 15 features has been selected using the recursive feature elimination technique. It has been found that these features provide significant discriminative information (p-value ≤ 0.05) for classification. On the test dataset of 82 EEG records, a classification accuracy, sensitivity, and specificity of 79.17%, 80.85%, and 76.00%, respectively have been achieved using a Quadratic Discriminant Analysis classifier. Results demonstrate that OSA exacerbation can be detected early and monitored with this simple yet effective method using a single channel EEG.