Automatic Detection of Obstructive Sleep Apnea Using Wavelet Transform and Entropy-Based Features From Single-Lead...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Sleep Medicine

Background:

• Obstructive sleep apnea (OSA) significantly impacts quality of life.
• Current gold standard, polysomnography, is time-consuming and costly.
• Electrocardiogram (ECG) signals offer a practical alternative for OSA detection.

Purpose of the Study:

• To propose a novel, automatic obstructive sleep apnea (OSA) detection method.
• To utilize single-lead ECG signals for OSA detection.
• To investigate nonlinear feature extraction and classification techniques.

Main Methods:

• Decomposition of ECG signals using wavelet transform (WT) with a Symlet function.
• Extraction of nonlinear and entropy-based features from WT coefficients.
• Feature selection using sequential forward selection and classification with Support Vector Machine (SVM).

Main Results:

• Achieved 94.63% accuracy for minute-by-minute classification and 95.71% for subject-by-subject classification.
• Entropy-based features demonstrated superior performance in extracting hidden ECG signal information.
• The method achieved high accuracy using only single-lead ECG signals.

Conclusions:

• The proposed method provides a highly accurate and automatic approach for OSA detection using single-lead ECG.
• Entropy-based features are effective for identifying OSA from ECG signals.
• The low computational load enables application in home monitoring systems.