Computer-aided obstructive sleep apnea screening from single-lead electrocardiogram using statistical and spectral features and bootstrap aggregating

Automatic sleep apnea screening is important to alleviate the onus of the physicians of analyzing a large volume of data visually. Again, the push towards low-power, portable and wearable sleep quality monitoring systems necessitates the use of minimum number of recording channels to enhance battery life. So, there is a dire need of an automated apnea detection scheme based on single-lead electrocardiogram (ECG). Most of the existing works are based on multiple channels of physiological signals or yield poor performance. The effect of various classification models on algorithmic performance is also poorly explored. In the present work, we propose a statistical and spectral feature based sleep apnea identification scheme that utilizes single-lead ECG signals. Bootstrap aggregating is employed to perform classification. The efficacy of the selected features is demonstrated by intuitive, statistical and graphical analyses. Optimal choices of classifier parameters are also expounded. The performance of the proposed algorithm is evaluated for various classifiers. The performance of our method is also compared to that of the state-of-the-art ones. The proposed method yields accuracy, sensitivity and specificity of 85.97%, 84.14% and 86.83% respectively on a widely used benchmark data-set. Experimental findings backed by statistical and graphical analyses suggest that the proposed method performs better than the existing ones in terms of accuracy, sensitivity, specificity and computational cost.