An automated method for sleep apnoea detection using HRV.

Area of Science:

• Biomedical Engineering
• Cardiology
• Sleep Medicine

Background:

• Sleep apnoea poses significant health risks, necessitating early and accurate diagnostic methods.
• Electrocardiogram (ECG) signals offer a non-invasive source for analyzing physiological changes during sleep.
• Heart Rate Variability (HRV) analysis derived from ECG can reveal autonomic nervous system dysregulation associated with apnoea events.

Purpose of the Study:

• To develop and validate a method for predicting the occurrence of respiratory apnoea events during sleep.
• To identify distinct HRV features indicative of impending apnoea episodes.
• To enhance the diagnostic accuracy of sleep apnoea detection using ECG signal processing.

Main Methods:

• Utilized ECG signals from 70 patients with sleep apnoea from the Physionet database.
• Generated HRV signals from ECG data and extracted time and frequency domain features.
• Employed statistical analysis, Principal Component Analysis (PCA), and a genetic algorithm for optimal feature selection.
• Evaluated feature performance individually and in combination for distinguishing between apnoea and non-apnoea intervals.

Main Results:

• Identified specific HRV features that effectively differentiate periods near apnoea events from those further away.
• Demonstrated that combining selected features significantly enhances the ability to detect apnoea events.
• Achieved high performance metrics: 99.77% specificity, 97.38% sensitivity, and 98.25% accuracy with the combined feature approach.

Conclusions:

• ECG-derived HRV analysis is a viable method for early detection of sleep apnoea.
• The proposed feature combination strategy offers superior diagnostic performance compared to previous studies.
• Early and accurate diagnosis of sleep apnoea can lead to timely interventions, improving patient outcomes and potentially saving lives.