Heart rate variability analysis for patients with obstructive sleep apnea

Obstructive sleep apnea (OSA) is a common health concern associated with serious implications and increased cardiovascular morbidity and mortality. This study approaches the problem of identification OSA patients and detection of OSA phases on the basis of heart rate variability (HRV) analysis. Only a single ECG-channel is required for this purpose. We used data from the apnea ECG database:40 patients with documented OSA and 20 controls divided into a learning and a test set of equal size. Commonly used HRV measures as well as some novel parameters are tested. The results are compared by ROC-analysis and promising parameters are combined into a multidimensional vector and evaluated by means of a second order polynomial classifier. Best results are obtained from parameters calculated by time delay embedding and correlation analysis of the interbeat interval series. For the identification task, 95% sensitivity and 100% specificity are achieved on the independent test set. The detection task yields an average classification rate of almost 85%.