Characterizing the NREM/REM sleep specific obstructive sleep apnea severity using snore sounds.

Area of Science:

• Medical Devices
• Sleep Medicine
• Biomedical Engineering

Background:

• Obstructive Sleep Apnea (OSA) diagnosis typically requires Polysomnography (PSG), a complex method with numerous sensors.
• Differentiating OSA severity between Rapid Eye Movement (REM) and non-REM (NREM) sleep states is clinically significant but challenging with current methods.
• Snoring, a common OSA symptom, is linked to upper airway (UA) muscle vibrations and distinct breathing patterns during different sleep stages.

Purpose of the Study:

• To investigate the feasibility of estimating Obstructive Sleep Apnea (OSA) severity during REM and NREM sleep states using only snoring and breathing sounds.
• To develop and validate a method for OSA severity assessment independent of Polysomnography (PSG).
• To explore the potential of using simple bedside sound acquisition devices for OSA monitoring.

Main Methods:

• Overnight audio recordings were collected from 91 patients undergoing PSG.
• Snore sounds were labeled according to REM/NREM sleep stages based on PSG data.
• Features capturing REM/NREM specific information from snore sounds were developed.
• Logistic regression (LR) classifier models were trained to map snore features to OSA severity bands, with separate models for males and females.

Main Results:

• High sensitivity (94-100%) and specificity (84-89%) were achieved for NREM stages at an OSA severity threshold of 30 events/h.
• High sensitivity (92-97%) and specificity (83-85%) were achieved for REM stages.
• Results demonstrate the feasibility of distinguishing severe/non-severe OSA in both REM and NREM sleep using only acoustic data.

Conclusions:

• Snore and breathing sounds contain sufficient information to estimate Obstructive Sleep Apnea (OSA) severity separately for REM and NREM sleep.
• Acoustic analysis of snoring offers a promising, non-invasive alternative to PSG for OSA severity assessment.
• Simple bedside devices, like mobile phones, could potentially be used for OSA monitoring based on this approach.