Contactless respiratory monitoring from a smartphone using pure tone suprasound

Respiratory monitoring during sleep is clinically valuable but existing approaches using contact-based sensors or dedicated hardware limit accessibility and longitudinal use. This study evaluated a smartphone-based system that emits a continuous pure tone suprasound signal (18–20 kHz) for contactless overnight respiratory monitoring, validated against a thoracic respiratory effort band (TREB) from polysomnography in 80 healthy participants. Respiratory signals were derived via in-phase/quadrature demodulation, with rate estimated by autocorrelation. A respiratory coherence score was developed to identify high-confidence segments without external reference, retaining 65% of sleep epochs. Within these epochs, respiratory rate agreement with the TREB yielded an ICC(2,1) of 0.97 and MAE of 0.29 brpm, and the derived signal showed a median absolute correlation of 0.85. Accuracy was comparable to Apple Watch estimates (ICC 0.95, MAE 0.46 brpm), with substantially greater temporal resolution. These results demonstrate that a consumer smartphone can contactlessly estimate respiratory rate during sleep with high accuracy.