Obstructive Sleep Apnea Syndrome Diagnosis using HRV Signal Processing

Diagnosing the sleep apnea syndrome is an important step toward sleep respiratory disorders. In this paper, an alternative, low-cost, reliable and effective system is proposed on the basis of HRV signal for classification. Different features are extracted from HRV signals, including time domain, frequency domain and time-frequency features. Here, ECG signals related to 10 patients from Physionet ECG database, are used. Moreover, Nonlinear Autoregressive Neural Network with external input (NARX) classifier is employed. Experimental results demonstrate that the sensitivity, specificity and accuracy rates are 93.3%, 91.8% and 92.55%, respectively. The high performance of the proposed system implies that extracted features from the HRV signal demonstrate a better diagnostic ability than other physiological signals and it can be used as an alternative for the PSG test to assist the physicians with the aim of improving the sleep apnea detection process.