SLUMBR: SLeep statUs estiMation from aBdominal Respiratory effort
Hector E. Romero, Ning Ma, Guy J. Brown, Sam R. Johnson · 2024
Accurately monitoring sleep for extended periods remains a challenge due to the cumbersome nature of conventional gold-standard techniques. We propose a novel deep learning method to estimate sleep status from an easily acquired abdominal respiratory effort signal. Our end-to-end convolutional neural network, developed on 476 hours of manually annotated polysomnography recordings from 53 participants, achieves an area under the curve of 0.90, and a more balanced performance across sensitivity and specificity than previous studies: 0.85 and 0.82, respectively. This method eliminates the need for obtrusive equipment and manual processing, paving the way for more accessible sleep monitoring solutions.