Development of Obstructive Sleep Apnea Event Detection Algorithms Based on Heart Rate Variability and ECG Morphology Features

  • 高 子平

Student thesis: Master's Thesis


Sleep medicine has become a salient issue in health and medical industry in the past decade This thesis proposes two electroencephalography (ECG) signal analysis algorithms for obstructive sleep apnea (OSA) detection The first algorithm is an ECG feature-based AdaBoost Bootstrap k-dimension tree k-nearest neighbor algorithm for OSA events recognition The proposed method processes single-lead ECG recordings to generate heart rate variability ECG-derived respiratory signals and cardiopulmonary coupling features for detecting the occurrence of sleep apnea and then provides a minute-by-minute analysis of disordered breathing The second algorithm is an ECG waveform detection method to locate the PQRST position of ECG signals After generating the ECG morphological features from the PQRST position a Classification and Regression Tree-based Random Forest algorithm was used to detect the OSA events The effectiveness and time consumption of the algorithms have been successfully validated by experimental results In the future we hope these algorithms can be applied to home care and obstructive sleep apnea early screening
Date of Award2015 Nov 16
Original languageEnglish
SupervisorJeen-Shing Wang (Supervisor)

Cite this