Potential of Electrocardiogram-Derived-Respiration based on QRS slopes and R-wave angle for Discriminating Apneic from Non-Apneic Segments

A first study on the potential of electrocardiogram(ECG)-derived respiration (EDR) signals based on QRS slopes and R-wave angles for sleep apnea detection is been presented. This EDR techniques have been previously validated with a wearable ECG armband for respiratory rate estimation. Furthermore, the amplitude of the oscillations in these EDR signals was observed to be related to the tidal volume in a previous pilot study. The hypothesis of this work is that that relation can be exploited for sleep apnea detection. A public data set (Physionet Apnea-ECG) composed of 105 polysomnography recordings was analyzed. A linear discriminant analysis was used using features related to the amplitude of the EDR oscillations. The classifier obtained an area under the curve from 0.74 to 0.82 when using the different analyzed features sets, suggesting that the relation between tidal volume and the amplitude of the EDR signals based on QRS slopes and R-wave angles can be exploited for sleep apnea detection. These results allow us to think of a sleep apnea screening tool based on the wearable armband and these EDR techniques, which would have both social and economic advantages.