Key result
Siamese-network transfer learning achieves a ~97% macro-F1 score for personalized AF detection from ECG segments.
Why the study?
Deep learning models require large amounts of annotated data to accurately detect atrial fibrillation, making it impractical to obtain sufficient personalized labels for individual patients.
Does a Siamese-network-based transfer learning approach accurately detect atrial fibrillation using minimal patient-specific single-lead ECG records?
Does a Siamese-network-based transfer learning approach accurately detect atrial fibrillation using minimal patient-specific single-lead ECG records?
A Siamese-network-based transfer learning approach using minimal patient-specific ECG data can achieve high accuracy (macro-F1 >96%) for personalized atrial fibrillation detection.
Answers are grounded in this paper and cite their sources.
Few-shot AF detectors may support personalized wearable screening; leaves open prospective validation before clinical adoption.
The proliferation of wearable devices has allowed the collection of electrocardiogram (ECG) recordings daily to monitor heart rhythm and rate. For example, 24-hour Holter monitors, cardiac patches, and smartwatches are widely used for ECG gathering and application. An automatic atrial fibrillation (AF) detector is required for timely ECG interpretation. Deep learning models can accurately identify AFs if large amounts of annotated data are available for model training. However, it is impractical to request sufficient labels for ECG recordings for an individual patient to train a personalized model. We propose a Siamese-network-based approach for transfer learning to address this issue. A pre-trained Siamese convolutional neural network is created by comparing two labeled ECG segments from the same patient. We sampled 30-second ECG segments with a 50% overlapping window from the ECG recordings of patients in the MIT-BIH Atrial Fibrillation Database. Subsequently, we independently detected the occurrence of AF in each patient in the Long-Term AF Database. By fine-tuning the model with the 1, 3, 5, 7, 9, or 11 ECG segments ranging from 30 to 180 s, our method achieved macro-F1 scores of 96.84%, 96.91%, 96.97%, 97.02%, 97.05%, and 97.07%, respectively.
No takes yet. Share an insight, caveat, or question.
Ng et al. (2023) studied Atrial fibrillation. Siamese-network-based transfer learning was evaluated on Macro-F1 score for atrial fibrillation detection. A Siamese-network-based transfer learning approach for personalized atrial fibrillation detection achieved a macro-F1 score of 97.07% when fine-tuned with 11 ECG segments.
Synapse has enriched 10 related papers. Consider them for comparative context: