ECG Signal-Based Feature Identification of Obstructive Sleep Apnea Using Dual-Tree Complex Wavelet Transform and Adaptive Neuro-Fuzzy Inference System

Obstructive sleep apnea (OSA) is a common disease that affects a large number of people around the world. It could be prevented if diagnosed early, but many cases go undetected due to the cost and limitations of home overnight polysomnography (PSG) systems. In this research, we propose a new diagnostic method for detecting the OSA disease using a feature extraction and classification approach from the electrocardiograph (ECG) signal only. The ECG can be easily extracted at home in a comfortable and minimally invasive way by anyone; then, the Dual-Tree Complex Wavelet Transform (DTCWT) is used to denoise the ECG signal. Five features were extracted, and the classification process was performed using the Adaptive Neuro Fuzzy Inference System (ANFIS).