Key result
Random Forest classifier achieves ~95% overall accuracy for automated ECG heartbeat classification.
Why the study?
Diagnosing ECG signals is challenging due to variability between patients and similarities in ECG patterns across different diseases, requiring improved feature extraction and classification methods.
Absolute Event Rate: 94.72% vs 91.87%
An automated ECG classification model using wavelet packet entropy and a Random Forest classifier achieved high accuracy in categorizing heartbeats, demonstrating potential for computer-aided arrhythmia detection.
ECG or electrocardiogram is an electrical signal which is generated by our heart. It is the cardiac electrical activity that provides important information about heart conditions [2]. ECG is very popular to identify heart illnesses like arrhythmia, chest pain, heart abnormalities, measuring heart rate, etc. In the past, till now ECG is the primary technique to detect heart illness in medical. ECG is a non-invasive technique. A survey World Health Organization says that heart diseases are the main reason for most deaths worldwide. In most cardiovascular diseases, arrhythmia is the most common. For this ECG is very much famous in medical studies. The study of an individual ECG beat can provide meaningfully correlated clinical information for the automatic ECG recognition of an ECG signal but it is difficult to investigate more ECG signals of different patients because of their different physical conditions. So here the main problem to investigating an ECG signal is that it can be different in every person. Suppose two different types of diseases have the same type of properties in an ECG signal. Even sometimes different patients have the same type of ECG pattern graph. These are the main difficulties in diagnosing an ECG signal. Many methods of feature extraction and classification have been proposed but some of the techniques remain to be improved. In this paper first of all we make our database with the help of the MIT-BIH database. After preprocessing and segmentation we decompose the signal by wavelet packet decomposition. Then calculate the entropy from the decomposed coefficients and extract the features.
No takes yet. Share an insight, caveat, or question.
Seba Maity (2023) studied Arrhythmia (n=17,207). Random Forest classifier with 36 features (wavelet packet entropy and statistical features) vs. Random Forest classifier with 32 features was evaluated on Overall classification accuracy. A Random Forest classifier using 36 features, including wavelet packet entropy and statistical features, achieved an overall accuracy of 94.72% for ECG heartbeat classification.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: