Efficient sleep apnea detection using single-lead ECG: A CNN-Transformer-LSTM approach.

Area of Science:

• Biomedical Engineering
• Cardiology
• Artificial Intelligence in Healthcare

Background:

• Sleep apnea (SA) is a common sleep disorder impacting respiratory patterns and potentially causing severe health complications.
• Early and accurate SA detection is crucial for preventing associated cardiac, cerebral, and pulmonary issues.
• Electrocardiograms (ECG) offer continuous heart monitoring, vital for identifying SA-related cardiac changes like arrhythmias.

Purpose of the Study:

• To develop and validate a hybrid neural network for automated sleep apnea detection using single-lead ECG signals.
• To assess the model's capability in capturing both spatial and temporal features for enhanced classification performance.
• To compare the model's efficacy against existing state-of-the-art methods for SA detection.

Main Methods:

• A hybrid CNN-Transformer-LSTM neural network model was designed for SA detection.
• The model processes RR intervals (RRI) and R-peak signals derived from ECG data.
• Performance was evaluated on the Physionet Apnea-ECG and UCDDB datasets using per-segment and per-recording classifications.

Main Results:

• The CNN-Transformer-LSTM model achieved 94.1% accuracy in per-segment classification (5-fold CV) on the Physionet dataset.
• Per-recording classification reached 100% accuracy with a 0.9996 correlation coefficient (5-fold CV).
• On the UCDDB dataset, accuracies of 99.37% (reduced) and 98.34% (full) were recorded, outperforming prior methods.

Conclusions:

• The CNN-Transformer-LSTM model demonstrates high effectiveness for sleep apnea detection from ECG.
• The model's performance suggests its potential utility in clinical and home-based SA screening devices.
• This approach offers a promising, non-invasive method for improved SA diagnosis and management.