Obstructive Sleep Apnea Detection Based on Sleep Sounds via Deep Learning

Abstract

This study aimed to propose a novel deep-learning method for automatic sleep apneic event detection and thus to estimate the apnea hypopnea index (AHI) and identify obstructive sleep apnea (OSA) in an event-by-event manner solely based on sleep sounds obtained by a noncontact audio recorder. Methods: We conducted a cross-sectional study of participants with habitual snoring or heavy breathing sounds during sleep to train and test a deep convolutional neural network named OSAnet for the detection of OSA based on sleep sounds. Polysomnography (PSG) was conducted, and sleep sounds were recorded simultaneously in a regular room without noise attenuation. The study was conducted in two phases. In phase one, eligible participants were enrolled and randomly allocated into training and validation groups for deep learning algorithm development. In phase two, eligible patients were enrolled in a test group for algorithm assessment. Sensitivity, specificity, accuracy, unweighted Cohen kappa coefficient (κ) and the area under the curve (AUC) were calculated using PSG as the reference standard. Results: A total of 135 participants were randomly divided into a training group (n, 116) and a validation group (n, 19). An independent test group of 59 participants was subsequently enrolled. Our algorithm achieved a precision of 0.81 and sensitivity of 0.78 in the test group for overall sleep event detection. The algorithm exhibited robust diagnostic performance to identify severe cases with a sensitivity of 95.6% and specificity of 91.6%. Conclusion: Our results showed that a deep learning algorithm based on sleep sounds recorded by a noncontact voice recorder served as a feasible tool for apneic event detection and OSA identification. This technique may hold promise for OSA assessment in the community in a relatively comfortable and low-cost manner. Further studies to develop a tool based on a home-based setting are warranted.

In-Home Smartphone-Based Prediction of Obstructive Sleep Apnea in Conjunction With Level 2 Home Polysomnography

“…Some studies focus on tracheal sounds, [24][25][26] while others use speech 28 or ambient sound. 8,9,27 However, many of these studies conducted sound recordings in controlled, noisefree environments, [26][27][28] which do not accurately represent the typical background noises experienced by patients in their home sleep settings. The ultimate goal of using sound-based approaches for OSA prediction is to develop a method that can be applied in a home setting, similar to level 1 in-laboratory PSG.…”

“…For example, a study used level 1 full-night PSG along with a noncontact digital voice recorder. 8 In that study, the level 1 PSG was conducted in a regular room without noise attenuation equipment to create a more realistic setting. However, it should be noted that all participants underwent the test in the same room within the hospital, which limits the generalizability to diverse real-world situations despite considering noise.…”

“…Sound-based assessment for OSA has emerged as a promising approach that enables noncontact and in-home monitoring using smartphones or wearable devices . Previous studies have demonstrated the potential of using breathing sounds to develop predictive models for OSA detection.…”

“…Sound-based assessment for OSA has emerged as a promising approach that enables noncontact and in-home monitoring using smartphones or wearable devices. [8][9][10] Previous studies have demonstrated the potential of using breathing sounds to develop predictive models for OSA detection. However, many of these studies validated their prediction models using sound collected in controlled environments, referencing level 1 PSG.…”

“…However, many of these studies validated their prediction models using sound collected in controlled environments, referencing level 1 PSG. 8,[10][11][12] Furthermore, previous studies failed to account for home environment noises, which are essential for accurate OSA prediction in a home setting. 10,12 To address this, our predeveloped model was trained using data that incorporated various home noises, ensuring its ability to accurately predict respiratory events in such environments.…”