Session P74.3

Screening Obstructive Sleep Apnoea Syndrome from Electrocardiogram Recordings Using Support Vector Machines

AH Khandoker*, CK Karmakar, M Palaniswami

The University of Melbourne
Melbourne, Australia

Obstructive sleep apnoea syndrome (OSAS), which is largely undiagnosed, is associated with cardiovascular morbidity as well as excessive daytime sleepiness and poor quality of life. In this study, we apply a machine learning technique [support vector machines (SVM)] for automated recognition of OSAS types from their nocturnal ECG recordings. Total 70 sets of nocturnal ECG recordings [35 data sets with apnoea (learning set) and 35 sets (test set)] acquired from normal subjects (OSAS-) and subjects with OSAS (OSAS+), each of approximately eight hours in duration, were downloaded from physionet and analysed. Features extracted from successive wavelet coefficient levels after wavelet decomposition of signals due to heart rate variability (HRV) from RR intervals and ECG derived respiration (EDR) from QRS amplitudes of whole record were presented as inputs to train the SVM mode to recognize OSAS+/- subjects. The optimal SVM parameter set (regularization and kernel parameters) was determined by using a leave-one-out procedure. Independent test results on 30 subjects showed that a SVM using a subset of selected combination of HRV and EDR features correctly recognized 20 out of 20 OSAS+ subjects and 10 out of 10 OSAS- subjects. For estimating the relative severity of OSAS, the posterior probabilities of SVM outputs were calculated and compared with respective apnoea index (apnea epochs/hour). These results suggest superior performance of SVM in OSAS recognition based on HRV and EDR features of ECG. These results demonstrate considerable potential in applying SVM in ECG based screening device that can aid sleep specialist in the initial assessment of patients with suspected OSAS.

(Abstract Control Number: 99)