Synapse
⌘+K
Synapse
PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
December 21, 2023

Convolutional Neural Network Based Deep Neural Network Model for Electrocardiogram Records Classification

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Manual analysis of large volumes of ECG data depletes medical resources, while conventional deep learning approaches face limitations such as manual characteristic identification, intricate models, and extensive training duration.

Does a 16-layer deep 1D convolutional neural network improve the classification of five types of heartbeats in ECG records?

Population

Five different types of heartbeats in the MIT-BIH Arrhythmia database

Design

Model development and validation study

Authors

AKAhmed Mahmood KhudhurMAMohammed Hasan AlwanQOQayssar Al Omairi

Discussion

Loading...

Member takes

Overview

May support automated ECG classification; leaves open prospective clinical validation.

Key Points

  • To develop a robust and resource-efficient deep 1D convolutional neural network capable of accurately classifying five distinct heartbeat classes from electrocardiogram recordings.
  • Utilized electrocardiogram data from the benchmark MIT-BIH Arrhythmia database to identify five distinct heartbeat categories.
  • Constructed a 16-layer one-dimensional convolutional neural network architecture divided into five functional groups, utilizing four groups for feature extraction and mapping alongside a final fully connected classification block.
  • Demonstrated superior classification metrics across overall accuracy, precision, and F1-score relative to conventional manual and intricate machine learning approaches.
  • Achieved low resource depletion during the training phase, confirming computational efficiency for clinical translation.

Structured PICO

Does a 16-layer deep 1D convolutional neural network improve the classification of five types of heartbeats in ECG records?

P
Population
ECG records from the MIT-BIH Arrhythmia database containing five different types of heartbeats
I
Intervention
16-layer deep one-dimensional convolutional neural network
O
Outcome
Classification performance (accuracy, precision, and F1-score)surrogate

A novel 16-layer 1D CNN model efficiently and accurately classifies five types of heartbeats from ECG data, potentially conserving medical resources.

Cite This Study

Khudhur et al. (2023) studied this question.

synapsesocial.com/papers/6a1a7c3e77ec05d9a7b89a7ahttps://doi.org/10.1145/3644713.3644862
View Full Paper
Ask AI
Bookmark
Share