Screening for obstructive sleep apnoea based on the electrocardiogram-the computers in cardiology challenge

摘要

The authors present a method of screening for obstructive sleep apnoea based on the electrocardiogram (ECG). The algorithm combines information from the ECG-derived respiration (EDR) signal and the RR interval tachogram. Power spectral features from the EDR signal were computed using the discrete harmonic wavelet transform, considering the power at the respiratory frequency and at frequencies below 0.1 Hz. Cycles of tachy/bradycardia (consistent with an arousal from sleep, as would be expected at the end of an episode of apnoea) were identified front the RR interval tachogram. Features were collated into minute-by-minute vectors and passed to a classifier. The algorithm correctly classified 81% of all minutes in the test database, with 29/30 patients correctly identified as apnoea or normal. Visual classification produced 92% correct classification, with all 30 patients correct.