Automated Detection of Sleep Apnea-Hypopnea Events Based on 60 GHz Frequency-Modulated Continuous-Wave Radar Using...

Area of Science:

• Biomedical Engineering
• Sleep Medicine
• Artificial Intelligence in Healthcare

Background:

• Obstructive sleep apnea (OSA) diagnosis relies on polysomnography (PSG), a resource-intensive gold standard.
• Non-contact sensing technologies offer a potential alternative for OSA screening and monitoring.
• Radar sensors present a promising non-contact method for physiological signal acquisition during sleep.

Purpose of the Study:

• To evaluate the feasibility of using 60 GHz frequency-modulated continuous-wave (FMCW) radar for automated detection of apnea-hypopnea events.
• To assess the performance of a convolutional recurrent neural network (CRNN) model for OSA diagnosis using radar data.
• To determine the correlation and agreement between radar-based estimations and ground truth PSG for OSA severity.

Main Methods:

• A dataset of 44 participants undergoing overnight PSG with an integrated radar sensor was utilized.
• A five-fold cross-validation approach was employed for model development and evaluation.
• CRNN models were trained to classify 1-minute sleep segments for apnea-hypopnea events.

Main Results:

• The area under the ROC curve for 1-minute segment classification ranged from 0.796 to 0.859.
• Sensitivities for apnea-hypopnea events were 49.0-67.6%, with 23.4-52.8 false positives per participant, varying by OSA severity.
• Strong correlations (Pearson's r = 0.805-0.949) and good to excellent agreement (ICC = 0.776-0.929) were found for the apnea-hypopnea index; substantial agreement (kappa = 0.648-0.736) for OSA severity.

Conclusions:

• Radar-based automated detection of breathing events is feasible for OSA diagnosis.
• The developed CRNN model shows potential for accurate OSA severity estimation using radar signals.
• Radar emerges as a promising standalone screening tool for obstructive sleep apnea.