Vibration signals of snoring as a simple severity predictor for obstructive sleep apnea

Background and Aim: Polysomnography (PSG), which involves simultaneous monitoring of various physiological monitors, is the current comprehensive tool for diagnosing obstructive sleep apnea (OSA). We aimed at validating vibrating signals of snoring as a single physiological parameter for screening and evaluating severity of OSA. Methods: Totally, 111 subjects from the sleep center of a tertiary referral center were categorized into four groups according to the apnea hypopnea index (AHI) obtained from PSG: simple snoring group (5 > AHI, healthy subjects, n = 11), mild OSA group (5 ≤ AHI < 15, n = 11), moderate OSA group (15 ≤ AHI < 30, n = 30) and severe OSA group (AHI ≥ 30, n = 59). Anthropometric parameters and sleep efficiency of all subjects were compared. Frequencies of amplitude changes of vibrating signals on anterior neck during sleep were analyzed to acquire a snoring burst index (SBI) using a novel algorithm. Data were compared with AHI and index of arterial oxygen saturation (Δ Index). Results: There were no significant differences in age and sleep efficiency among all groups. Bland–Altman analysis showed better agreement between SBI and AHI (r = 0.906, P < 0.001) than Δ Index and AHI (r = 0.859, P < 0.001). Additionally, receiver operating characteristic (ROC) showed substantially stronger sensitivity and specificity of SBI in distinguishing between patients with moderate and severe OSA compared with Δ Index (sensitivity: 81.4% vs 66.4%; specificity: 96.7% vs 86.7%, for SBI and Δ Index, respectively). Conclusion: SBI may serve as a portable tool for screening patients and assessing OSA severity in a non-hospital setting.