Apnea-hypopnea index (AHI) estimation using breathing Sounds, accelerometer and pulse oximeter

Sleep apnea (SA) severity is determined by the apnea-hypopnea index (AHI). 85% of patients remain undiagnosed due to the complexity and limited access to polysomnography (PSG). Accordingly, a system for convenient and reliable diagnosis of SA is urgently needed. To address this problem, we aim to develop a wearable technology for AHI estimation using breathing sounds, respiratory related movement and blood oxygen saturation (SaO₂). Adults with suspected SA underwent overnight PSG at Toronto Rehabilitations Institute. During PSG, breathing sounds and respiratory related movements were recorded over the suprasternal notch using a microphone and an accelerometer, respectively (the Patch (Figure 1-a)). At first, we calculated the amount of drops in SaO₂ from pulse oximeter. Subsequently, energy of breaths and accelerometer were extracted. Features were a) normalized, b) weighted, and c) summed. An event was marked when the summed score was greater than a threshold. Finally, AHI was estimated by dividing the number of events by recording time. Estimated AHI (Patch_AHI) was compared to the AHI obtained from PSG (PSG_AHI). Data from 50 subjects were investigated. A high correlation was observed between the Patch_AHI and PSG_AHI (Figure 1-b, r² = 0.93). Considering a diagnostic cut-off AHI of =15, sensitivity and specificity of 92.3% and 91.8% were achieved, respectively. Our proposed system is easy to operate and comfortable, and it is currently going through clinical trials for final validation.