Abstract

Snoring sounds vary significantly within and between snorers. In this study, the variation of snoring sounds and its association with obstructive sleep apnea (OSA) are quantified. Snoring sounds of 42 snorers with different degrees of obstructive sleep apnea and 15 non-OSA snorers were analyzed. The sounds were recorded by a microphone placed over the suprasternal notch of trachea, simultaneously with polysomnography (PSG) data over the entire night. We hypothesize that snoring sounds vary significantly within a subject depending on the level of obstruction, and thus the level of airflow. We also hypothesize that this variability is associated with the severity of OSA. For each individual, we extracted snoring sound segments from the respiratory recordings, and divided them into three classes: non-apneic, hypopneic, and post-apneic using their PSG information. Several features were extracted from the snoring sound segments, and compared using a nonparametric statistical test. The results show significant shift in the median of features among the snoring sound classes (p<0.00001) of an individual. In contrast to hypopneic and post-apneic classes, the characteristics of snoring sounds did not vary significantly over time in non-apneic class. Therefore, we used the total variation norm of each subject to classify the participants as OSA and non-OSA snorers. The results showed 92.9% sensitivity, 100% specificity and 96.4% accuracy.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call