PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
Synapse
⌘+K
Synapse
April 25, 2023PLoS ONEOpen Access

ECG classification using 1-D convolutional deep residual neural network

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Traditional ECG classification methods rely on complex signal processing phases that lead to expensive designs.

Does a 1-D convolutional deep residual neural network with SMOTE improve the classification accuracy of ECG heartbeats?

Population

ECG signals from the PhysioNet MIT-BIH Arrhythmia database

Comparison

1-D convolutional deep ResNet with SMOTE vs other 1-D CNNs

Design

Algorithm development and ten-fold cross validation study

Key result

The proposed 1-D convolutional deep residual neural network with SMOTE achieved an average accuracy of 98.63% for the classification of five heartbeat types.

Authors

FKFahad KhanXYXiaojun YuZYZhaohui Yuan

Discussion

Loading...

Member takes

Overview

May support automated ECG analysis; hypothesis-generating pending external clinical validation.

Structured PICO

Does a 1-D convolutional deep residual neural network with SMOTE improve the classification accuracy of ECG heartbeats?

P
Population
109,446 heartbeats from 44 records (derived from 47 individuals) in the PhysioNet MIT-BIH Arrhythmia database
I
Intervention
1-D convolutional deep residual neural network (ResNet) combined with synthetic minority oversampling technique (SMOTE)
C
Comparator
Plain networks and other 1-D CNNs (implicit/historical)
O
Outcome
Classification accuracy, precision, sensitivity, specificity, F1-score, and kappa for five heartbeat types (N, S, V, F, Q)surrogate

Main Result

Absolute Event Rate: 98.63% vs 95%

A 1-D convolutional deep residual neural network combined with SMOTE for data balancing achieves high accuracy (98.63%) in classifying five types of ECG heartbeats.

Cite This Study

Khan et al. (2023) studied Arrhythmia (n=47). 1-D convolutional deep residual neural network (ResNet) with SMOTE vs. Model without SMOTE was evaluated on Classification accuracy. The proposed 1-D convolutional deep residual neural network with SMOTE achieved an average accuracy of 98.63% for the classification of five heartbeat types.

synapsesocial.com/papers/6a1588435347fbb1739fece5https://doi.org/10.1371/journal.pone.0284791
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1ECG signal classification based on CNN deep learning2023 · 5 citations
  2. 2Classification of ECG signals using deep neural networks2023 · 9 citations
  3. 3Classification of ECG signals based on 1D convolution neural network2017 · 238 citations
  4. 4Convolutional Neural Network Based Deep Neural Network Model for Electrocardiogram Records Classification2023 · 1 citations
  5. 5Improving ECG Classification Performance by Using an Optimized One-Dimensional Residual Network Model2022 · 2 citations