Non-Contact Sleep Apnea Detection System Based on Data-Driven Methods Using FMCW Radar

Sleep apnea is a prevalent and serious sleep disorder. Conventional diagnostic methods mainly rely on contact sensors, which are inconvenient and uncomfortable. Radar-based non-contact sleep monitoring overcomes these drawbacks and can serve as an effective alternative. This paper proposes a non-contact sleep apnea detection system based on FMCW radar using machine learning. Feature extraction and deep learning methods established the correlation between respiratory signals and abnormal events. The feature-based approach achieved 88.42% accuracy, 0.8700 sensitivity, 0.8935 specificity, and 0.7592 Kappa in 23 clinical data cases.