Key result
The 8CSL model combining CNN, LSTM, and shortcut connections achieved an average test F1 score of up to 89.55% for detecting atrial fibrillation from 10-second ECG segments.
Why the study?
Early diagnosis of atrial fibrillation can improve treatment effectiveness and prevent serious complications, motivating automatic ECG classification models.
Does the 8CSL deep learning model improve the automatic detection of atrial fibrillation from ECG recordings compared to RNN and MCNN?
Population
ECG recordings from the Computing in Cardiology Challenge 2017 dataset
Comparison
8CSL model vs RNN and MCNN
Design
Model development and 10-fold cross-validation study
Authors
Loading...
8CSL CNN-LSTM detects AF on ECG; leaves open prospective validation before clinical use.
Does the 8CSL deep learning model improve the automatic detection of atrial fibrillation from ECG recordings compared to RNN and MCNN?
The proposed 8CSL deep learning model demonstrates high accuracy (F1 score up to 89.55%) for the automatic detection of atrial fibrillation from short ECG segments.
Ping et al. (2020) studied Atrial fibrillation. 8CSL (8-layer CNN with shortcut connection and 1-layer LSTM) vs. Recurrent neural networks (RNN) and multi-scale convolution neural networks (MCNN) was evaluated on F1 score for ECG classification. The 8CSL model combining CNN, LSTM, and shortcut connections achieved an average test F1 score of up to 89.55% for detecting atrial fibrillation from 10-second ECG segments.
Synapse has enriched 3 closely related papers on similar clinical questions. Consider them for comparative context: