What is the clinical evidence from this study?

Study design: Other. Population: Arrhythmia (n=32422). Intervention: 1D 12-layer Convolutional Neural Network vs. BP neural network, Random Forest, and benchmarked CNN. Primary outcome: Overall classification accuracy (p=0.0001).

January 5, 2021Open Access

A Study on Arrhythmia via ECG Signal Classification Using the Convolutional Neural Network

Key Result

The proposed 1D 12-layer convolutional neural network achieved an overall classification accuracy of 97.41% for five micro-classes of heartbeats, significantly outperforming traditional machine learning models.

Structured PICO

Does a 12-layer 1D-CNN improve the classification accuracy of 5 micro-classes of heartbeats in the MIT-BIH Arrhythmia database compared to traditional machine learning methods?

Population

32,422 heartbeats extracted from 16 recordings of the MIT-BIH Arrhythmia database, classifying 5 micro-classes (Normal, Left Bundle Branch Block, Right Bundle Branch Block, Atrial Premature, Premature Ventricular Contraction)

Intervention

12-layer deep one-dimensional convolutional neural network (1D-CNN) with wavelet self-adaptive threshold denoising

Comparator

BP neural network, Random Forest network, and a benchmarked CNN network

Outcome

Classification accuracy, sensitivity, specificity, and positive prediction ratesurrogate

A 12-layer 1D-CNN effectively classifies five micro-classes of heartbeats with 97.41% accuracy, outperforming traditional machine learning models and demonstrating strong anti-noise capabilities.

Main Result

Absolute Event Rate: 97.41% vs 95.72%

p-value: p=0.0001

Limitations

Computationally intensive to train the network
Have yet to apply any optimization techniques to optimize the model parameters

Abstract

Cardiovascular diseases (CVDs) are the leading cause of death today. The current identification method of the diseases is analyzing the Electrocardiogram (ECG), which is a medical monitoring technology recording cardiac activity. Unfortunately, looking for experts to analyze a large amount of ECG data consumes too many medical resources. Therefore, the method of identifying ECG characteristics based on machine learning has gradually become prevalent. However, there are some drawbacks to these typical methods, requiring manual feature recognition, complex models, and long training time. This paper proposes a robust and efficient 12-layer deep one-dimensional convolutional neural network on classifying the five micro-classes of heartbeat types in the MIT- BIH Arrhythmia database. The five types of heartbeat features are classified, and wavelet self-adaptive threshold denoising method is used in the experiments. Compared with BP neural network, random forest, and other CNN networks, the results show that the model proposed in this paper has better performance in accuracy, sensitivity, robustness, and anti-noise capability. Its accurate classification effectively saves medical resources, which has a positive effect on clinical practice.

Bookmark

View Full Paper