Key result
The proposed deep learning models for automated feature extraction from ECG signals achieved an accuracy of 99.92% for fall detection and 99.93% for myocardial infarction classification.
Why the study?
Most deep learning methods for automatic ECG classification rely on complex models or manually engineered features based on domain knowledge, under the assumption that deeper and more complex models learn better.
Population
Three publicly available datasets, including the PTB database
Comparison
CNN-LSTM hybrid and attention/transformer-based models vs contemporary deep neural networks
Design
Model development and validation study
Authors
Loading...
Simplifies DL for ECG classification; leaves open prospective validation before clinical adoption.
The proposed deep learning models for automated feature extraction from ECG signals achieved high classification accuracy for fall detection and myocardial infarction, comparable to or exceeding contemporary complex models.
Butt et al. (2022) studied Electrocardiogram (ECG) signal classification (fall detection and myocardial infarction). CNN-LSTM hybrid model and attention/transformer-based model with wavelet transform vs. Contemporary deep neural networks was evaluated on Classification accuracy. The proposed deep learning models for automated feature extraction from ECG signals achieved an accuracy of 99.92% for fall detection and 99.93% for myocardial infarction classification.
Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context: