Key result
Deep neural network methods such as CNN and RNN provide high accuracy for ECG signal classification by automatically extracting features, outperforming conventional methods.
Neural network methods like CNN and RNN provide the best accuracy for ECG signal classification.
Supports CNN/RNN testing in ECG research; extends ML literature but leaves open clinical outcome validation.
The heart is an important part of the human body, functioning to pump blood through the circulatory system. Heartbeats generate a signal called an ECG signal. ECG signals or electrocardiogram signals are basic raw signals to identify and classify heart function based on heart rate. Its main task is to analyze each signal in the heart, whether normal or abnormal. This paper discusses some of the classification methods which most frequently used to classify ECG signals. These methods include pre-processing, feature extraction, and classification methods such as MLP, K-NN, SVM, CNN, and RNN. There were two stages of ECG classification, the feature extraction stage and the classification stage. Before ECG features were extracted, raw ECG signal data first processed in the pre-processing stage because ECG signals were not necessarily free of noise. Noise will cause a decrease in accuracy during the classification process. After features were extracted, ECG signals were then classified with the classification method. Neural Network methods such as CNN and RNN are best to use since they can give better accuracy. For further research, the machine learning method needs to be improved to get high accuracy and high precision in the ECG signals classification.
No takes yet. Share an insight, caveat, or question.
Fikri et al. (2021) conducted a review in Arrhythmia. Deep neural networks (CNN and RNN) vs. Conventional machine learning methods was evaluated on Classification accuracy. Deep neural network methods such as CNN and RNN provide high accuracy for ECG signal classification by automatically extracting features, outperforming conventional methods.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: