Sleep apnea classification using least-squares support vector machines on single lead ECG

Area of Science:

• Biomedical Engineering
• Cardiology
• Sleep Medicine

Background:

• Sleep apnea is a common disorder with significant health implications.
• Accurate detection of sleep apnea events is crucial for timely diagnosis and treatment.
• Existing methods for sleep apnea detection can be complex or resource-intensive.

Purpose of the Study:

• To present a novel methodology for identifying sleep apnea events.
• To develop a computationally simple yet accurate detection system.
• To enable effective sleep apnea monitoring in resource-limited settings like home environments.

Main Methods:

• Utilized four easily computable features: two from RR interval time series, one from an approximate respiratory signal derived via principal component analysis (PCA) from ECG.
• Introduced a fourth feature computed from the principal components of QRS complexes.
• Employed a least squares support vector machine (LS-SVM) classifier with a Radial Basis Function (RBF) kernel.

Main Results:

• Achieved >85% accuracy for subject-independent sleep apnea classification.
• Attained >90% accuracy for patient-specific sleep apnea classification.
• Demonstrated comparable performance to existing literature methods with significantly simpler computation.

Conclusions:

• The proposed methodology offers a straightforward and efficient approach to sleep apnea event identification.
• The method's simplicity makes it suitable for implementation in home monitoring systems with limited computational power.
• High accuracy in both subject-independent and patient-specific classifications validates the proposed features and classifier.