Key result
ECG-based machine learning achieves up to ~99% accuracy for classifying arrhythmias and CHF.
Why the study?
Early detection and prevention of cardiovascular diseases using ECG is challenging due to the non-stationary and variable nature of ECG signals, making manual observation time-consuming and potentially misleading.
Does a machine learning approach using morphological ECG features accurately detect myocardial infarction and dilated cardiomyopathy?
Does a machine learning approach using morphological ECG features accurately detect myocardial infarction and dilated cardiomyopathy?
A machine learning framework using morphological ECG features demonstrates high accuracy in detecting myocardial infarction and dilated cardiomyopathy from standard datasets.
May aid automated ECG screening for MI and DCM; hypothesis-generating and requires prospective clinical validation.
Cardiovascular Diseases (CVD’s) forms the vital basis of death in the world population. The human mortality rate due to CVD’s can be reduced with early detection and prevention of cardiac threat. Electrocardiogram (ECG) indicates the state of the cardiac heart and thus, acts as a standard to the cardiac health of an individual. ECG is a non-invasive, clinical therapeutic agent, which can be used to detect CVD in early stage and prognosis of threaten cardiac diseases. Since, the ECG is an abruptly changing, non-stationary waveform; the anomalies are non-periodic and may appear at varied fluctuations and intervals on random scale. Manual observation of ECG can be hectic in time and may be misleading to medical experts. So, it is necessary to endorse a computer assist framework for rapid and precise diagnosis of CVD’s. This paper deals with the automated characterization and detection of myocardial infarction (MI), dilated cardiomyopathy (DCM) diseases using morphological features extracted from ECG signal and machine learning approach. Our proposed approach attained maximum classification accuracy 87.5% for MIT-BIH Arrhythmia dataset and 99.33% accuracy for BIDMC Congestive Heart Failure (CHF) dataset with K-nearest neighbor (kNN) clustering and decision tree classifier with entropy criterion. The clinical experts can use our methodology to make precise prognosis of mentioned CVDs. Hence, survival prospect of a patient can be notably enlarged by prompt detection and clinical therapy of CVDs.
No takes yet. Share an insight, caveat, or question.
Karnewar et al. (2023) studied Myocardial infarction, dilated cardiomyopathy, congestive heart failure. Automated characterization and detection using ECG morphological features and machine learning was evaluated on Classification accuracy. An automated machine learning approach using ECG morphological features achieved a maximum classification accuracy of 87.5% for the MIT-BIH Arrhythmia dataset and 99.33% for the BIDMC CHF dataset.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: