Obstructive sleep apnea screening by heart rate variability-based apnea/normal respiration discriminant model.

摘要: Objective Obstructive sleep apnea (OSA) is a common disorder; however, most patients are undiagnosed and untreated because it difficult for themselves to notice OSA in daily living. Polysomnography (PSG), which the gold standard test disorder diagnosis, cannot be performed many hospitals. This fact motivates us develop simple system screening at home. Approach The autonomic nervous changes during apnea, such affect heart rate variability (HRV). work develops new method based on HRV analysis machine learning technologies. An apnea/normal respiration (A/N) discriminant model built condition estimation every measurement, an apnea/sleep ratio introduced final diagnosis. A random forest adopted A/N construction, trained with PhysioNet apnea-ECG database. Main results performance of proposed was evaluated by applying clinical PSG data. Sensitivity specificity achieved 76% 92%, respectively, comparable existing portable monitoring devices used laboratories. Significance Since can more easily than devices, will contribute treatment.