| Background. Electrocardiography (ECG) based methods has been proposed as an alternative obstructive sleep apnea (OSA) screening tool in adults to the standard sleep study polysomnography (PSG) due to its high cost and inconvenience. For pediatric population, ECG-based approach for OSA may be more suitable because of their less complicated cardiovascular conditions compared with adults. The success of such screening tool could improve the pediatric management after adenoids-tonsillectomy (A&T), the clinical treatment for pediatric OSA. This study explored and evaluated a novel method for screening pediatric OSA severity based on Electrocardiography.; Method. The nocturnal cardio-respiratory interaction among pediatric patients was simplified as two possible distinguished regular patterns, respiratory sinus arrhythmia (RSA) and cyclic variation of heart rate (CVHR), a heart rate signature of repeated OSA breathing. Overnight heart rate patterns were analogized as sentences that are consist of regular components (words) and meaningful transient events (comma, period, etc.). In such "lexon" model for nocturnal heart rate, these components are RSA and CVHR while transient event indicates diminish of RSA due to any non-periodic reasons (arousals, moving, etc). A set of cardio-respiratory indices (CRI) was developed to stratify and identify these components based on the autocorrelation function of heart beat intervals (HBI). CRI estimates the mean and standard deviation of a locking number of heart beats within each respiratory event as well as the mean respiratory rate. A short-time analysis of CRI was applied on each patient's overnight HBI to identify RSA region and CVHR region. Apnea and hypopnea index (AHI) was estimated on the CVHR detection to decide the severity of OSA. Such decision was evaluated by using clinical diagnosis as standards.; Results. A total of 105 pediatric PSG were studied retrospectively. Case to case OSA detection reached sensitivity (SS) of 75%, specificity (SP) of 90%, positive predictive value (PPV) of 69%, and negative predictive value (NPV) of 92% respectively. SS, SP, PVV, and NPV were improved to 100%, 85%, 67%, and 100% respectively with the assistance of arterial oxygen saturation from PSG. |
