Key result
LSTM model achieves ~99% accuracy classifying ECGs into arrhythmia, normal sinus rhythm, and congestive failure.
Why the study?
Accurately detecting cardiac arrhythmia in clinical settings is difficult due to noise and overlapping acute and chronic cardiac issues.
Does a deep learning neural network (LSTM) improve the classification accuracy of ECG recordings for detecting cardiac arrhythmias compared to state-of-the-art techniques?
Does a deep learning neural network (LSTM) improve the classification accuracy of ECG recordings for detecting cardiac arrhythmias compared to state-of-the-art techniques?
A deep learning LSTM model with four hidden layers and Adam optimization achieved 99.26% accuracy in classifying ECGs into normal sinus rhythm, arrhythmia, and congestive heart failure.
May aid automated ECG classification; leaves open prospective validation and SOTA comparisons before clinical use.
The process of detecting the cardiac arrhythmia accurately for the subjects who were kept under the clinical arrangements are very difficult and critical because the observation readings normally produce reflections over chronic cardiac associated problems from the acute problems. There are certain variable factors visualized that are existing between the clinical subjects and this exists because of certain types of noises which could not be avoided. This could be a tedious problem even for some well experienced clinical experts. In this proposed study, the exploration is made towards the design and implementation of deep learning neural networks for the purpose of making the classification towards the recording of ECG using the residual and recurrent architectures. The testing towards the proposed methodology has been made associated with the datasets of ECG comprising 170 subjects' data that possess triplet classes inclusive of abnormal cardiac rhythm (Arrhythmia), Normal Sinus Rhythm and congestive failure of cardiac muscle. Attained results has proven that the proposed methodology of LSTM which possess quad hidden layers and the Adam optimization function produces the peak accuracy of 99.26% and it is inferred to be very higher value of accuracy when compared with state of art techniques.
No takes yet. Share an insight, caveat, or question.
Alagarsamy et al. (2022) studied Cardiac arrhythmia and congestive heart failure (n=170). LSTM deep learning neural network with quad hidden layers and Adam optimization vs. State of art techniques was evaluated on Classification accuracy. An LSTM deep learning model with quad hidden layers and Adam optimization achieved a peak accuracy of 99.26% in classifying ECG recordings into arrhythmia, normal sinus rhythm, and congestive failure.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: