Sleep-Wake Evaluation from Whole-Night Non-Contact Audio Recordings of Breathing Sounds

Abstract

Study ObjectivesTo develop and validate a novel non-contact system for whole-night sleep evaluation using breathing sounds analysis (BSA).DesignWhole-night breathing sounds (using ambient microphone) and polysomnography (PSG) were simultaneously collected at a sleep laboratory (mean recording time 7.1 hours). A set of acoustic features quantifying breathing pattern were developed to distinguish between sleep and wake epochs (30 sec segments). Epochs (n = 59,108 design study and n = 68,560 validation study) were classified using AdaBoost classifier and validated epoch-by-epoch for sensitivity, specificity, positive and negative predictive values, accuracy, and Cohen's kappa. Sleep quality parameters were calculated based on the sleep/wake classifications and compared with PSG for validity.SettingUniversity affiliated sleep-wake disorder center and biomedical signal processing laboratory.PatientsOne hundred and fifty patients (age 54.0±14.8 years, BMI 31.6±5.5 kg/m2, m/f 97/53) referred for PSG were prospectively and consecutively recruited. The system was trained (design study) on 80 subjects; validation study was blindly performed on the additional 70 subjects.Measurements and ResultsEpoch-by-epoch accuracy rate for the validation study was 83.3% with sensitivity of 92.2% (sleep as sleep), specificity of 56.6% (awake as awake), and Cohen's kappa of 0.508. Comparing sleep quality parameters of BSA and PSG demonstrate average error of sleep latency, total sleep time, wake after sleep onset, and sleep efficiency of 16.6 min, 35.8 min, and 29.6 min, and 8%, respectively.ConclusionsThis study provides evidence that sleep-wake activity and sleep quality parameters can be reliably estimated solely using breathing sound analysis. This study highlights the potential of this innovative approach to measure sleep in research and clinical circumstances.

Breathing and Snoring Sound Characteristics during Sleep in Adults

“…This finding confirms early reports showing that snoring occurs mainly during inspiration, and that expiratory sound events are not common. 9,10,21,25 During sleep, upper airway resistance increases, 1-7 leading to amplification of air-pressure oscillations that are perceived as typical breathing sounds. 10,25 These sounds can vary from quiet (< 40 dB) to very loud snoring in the same subject.…”

“…9,10,21,25 During sleep, upper airway resistance increases, 1-7 leading to amplification of air-pressure oscillations that are perceived as typical breathing sounds. 10,25 These sounds can vary from quiet (< 40 dB) to very loud snoring in the same subject. 10 We found that, on average, 45% of the respiratory events during sleep generated air pressure oscillations that could be detected and analyzed.…”

“…45 Using an audio-based approach, it is important to improve signal-to-noise ratio in order to capture the signals of interest (breathing and snoring sounds). To achieve this, we used an adaptive spectral subtraction technique that subtracted the estimated background noise, 10,25 which has minimal distortion effect on sound intensity.…”

“…10 We also found that more than 97% of the snoring events occurred during inspiration; expiratory sound events are not common, confirming earlier reports. 9,21 This approach of detection of breathing sounds was further validated to reliably estimate apnea-hypopnea index (AHI) 24 and sleep-wake activity 25 in laboratory and at-home environments. It is commonly known that snoring intensity (dB) is higher for men than for women 13,21 and it is correlated with AHI.…”

Noncontact Sleep Study by Multi-Modal Sensor Fusion

“…The commercialized product ResMed S+ classifies sleep stages using multiple sensors and a complex algorithm that is not openly available. By combining the ideas of Tauhidur Rahman et al [ 11 ], Eliran Dfana et al [ 15 ] and ResMed [ 16 ], we suggest that the fusion of noncontact microphone sensor and radar sensor data-based advanced signal processing algorithms would likely enhance the accuracy of sleep stage classification. Moreover, most previously-proposed sleep stage classification methods are aimed at average users, not patients with sleep disorders who actually require sleep stage information.…”