Prediction of sleep apnea episodes from a wireless wearable multisensor suite

摘要: Obstructive sleep apnea (OSA) is a common disorder found in 24% of adult men and 9% women. Although continuous positive airway pressure (CPAP) has emerged as standard therapy for OSA, majority patients are not tolerant to this treatment due the uncomfortable nasal air delivery during their sleep. We introduce Dirichlet process Gaussian Mixture (DPGM) model predict occurrence episodes based on analyzing complex cardiorespiratory signals gathered from custom-designed wireless wearable multisensory suite. Extensive testing with PhysioNet's OSA database suggests that accuracy offline classification 88%. Accuracies predicting an episode 1-min ahead 83% 3-min 77%. Such accurate prediction impending can be used adaptively adjust CPAP airflow (towards improving patients' adherence), or torso posture (e.g., minor chin adjustments maintain steady levels airflow).