Parametric Power Spectrum Analysis of ECG Signals for Obstructive Sleep Apnoea Classification

Abstract

This work applies time-varying parametric power spectral density analysis to ECG and derived signals in order to discover the frequency components related to obstructive sleep apnoea. Heart rate variability signals were derived from the original ECG signals using R-R wave intervals. The power spectral densities were calculated using a parametric method across the heart rate variability frequency bands. Based on the power spectrum values, a number of beat-by-beat frequency power features were extracted from a PhysioNet dataset and studied together with the PhysioNet apnoea expert annotations.