Prediction of sleep apnea episodes from a wireless wearable multisensor suite

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