Detection of obstructive sleep apnea in children using decreases in the amplitude fluctuations of PPG signal and HRV

An analysis of the HRV during decreases in the amplitude fluctuations of PPG (DAP) events, and their utility in OSAS screening is presented. The overall data set used in the study includes the polysomnographic records of 21 children. DAP events were automatically detected by an algorithm based on the envelope attenuations of the PPG. DAP events were classified as apneic or non apneic by a linear discriminant analysis. The features used by the linear discriminant come from the temporal and spectral parameters of the heart rate obtained by Smooth Pseudo Wigner Ville Distribution. Two indexes were defined: the number of DAP events per hour ratio rDAP and the number of apneic DAP events per hour ratio rDAP^α. Results show a 12% increase in accuracy for rDAP^α with respect to rDAP in classifying 1 hour polysomnographic segments, reaching values of 72.7% and 80% for sensitivity and specificity, respectively. As for subject classification, the improvement in accuracy is 6.7% obtaining values of 87.5% and 71.4% for sensitivity and specificity respectively. These results suggest that the combination of DAP and HRV could be an alternative for sleep apnea screening with the added benefit of low cost and simplicity.