[An algorithm based on ECG signal for sleep apnea syndrome detection].

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Context:

• Sleep apnea syndrome (SAS) diagnosis is crucial for preventing cardiovascular and cerebrovascular disorders.
• Current diagnostic methods can be invasive or require specialized equipment.
• Electrocardiogram (ECG) signals offer a non-invasive and readily available data source.

Purpose:

• To develop a novel, non-invasive method for sleep apnea syndrome detection.
• To utilize single-channel ECG signals for SAS diagnosis.
• To evaluate the efficacy of a machine learning approach for classifying sleep apnea events.

Summary:

• A novel method for sleep apnea syndrome (SAS) detection was developed using single-channel electrocardiogram (ECG) signals.
• ECG signals were preprocessed to extract RR and ECG-derived respiratory (EDR) signals.
• Support vector machine (SVM) classification, utilizing 40 time- and spectral-domain features, achieved 95% accuracy on training sets and 88% on test sets using the MIT-BIH Apnea-ECG database.

Impact:

• Provides a potentially simpler and more accessible method for SAS diagnosis.
• Highlights the utility of ECG-based analysis for respiratory disorder detection.
• Contributes to the early identification and management of SAS, mitigating associated health risks.