An electrocardiography system design for obstructive sleep apnea detection based on improved lomb frequency analysis algorithm

Wai-Chi Fang, I-Wei Chen, Shu-Han Fan, Chih-Kuo Lee

Abstract: In this paper, we present a real-time method for obstructive sleep apnea (OSA) detection of frequency analysis of ECG-derived respiratory (EDR) and heart rate variability (HRV). The method is computationally simple with ECG signals to determine the time interval of OSA. We compare it to a traditional complexly Polysomnography (PSG) which needs several physiological signals measured from patients. For the wearable real-time application, the simplified Lomb Periodogram is proposed to perform the frequency analysis of EDR and HRV. The data from 900 ECG recordings from MIT PhysioNet Sleep Apnea database was utilized in the paper. The approximated method of OSA method obtained the highest Specificity (Sp) 90.1%, Sensitivity (Se) 94.3%, and Accuracy 92.1%.