Automated Prediction and Occurence of Tachyarrhythmia using Deep learning Techniques

• Published in: 2022 Third International Conference on Intelligent Computing Instrumentation and Control Technologies (ICICICT) pp. 1230 - 1235
• Main Authors: Naresh, D., Reddy, G. Himaja, Kondiparthi, Tharuni, Jashmika, V. S., Kumari, Ch. Usha
• Format: Conference Proceeding
• Language: English
• Published: IEEE 11.08.2022
• Subjects: Cardiac arrest; Convolutional Neural Network (CNN); Convolutional neural networks; Deep learning; ECG-Electrocardiogram (ECG); Electrocardiography; High Rate Variability (HRV); Implantable Cardioverter Defibrillators; Shape; Signal processing algorithms; Spatial Fuzzy C-Means (SFCM); Transfer learning; Ventricular Fibrillations(VF); Ventricular Tachyarrhythmia (VTA)
• DOI: 10.1109/ICICICT54557.2022.9917672

Cardiac arrest is the top global cause of death. Diagnosis and treatment is too late due to lack of awareness among the people. Tachyarrhythmia is the abnormal heart rate which may lead to cardiac arrest in the worst cases. This project aims to detect, predict and find the occurrence of Ventricular Tachyarrhythmia using Neural Networks such as CNN and DNN. Convolutional Neural Networks(CNN) were prepared and tried with two competences, HRV and QRS complex shape, for cross validation. The prospective model highlights the High Rate Variability(HRV) which is extricated from electrocardiograms (ECGs) as the contribution to prepare AI calculations. High Rate Variability(HRV) pulses are time variants with two progressive QRS buildings (Q, R, and S waves in ECG). Deep learning calculations utilize a foreordained coherent architecture to break down information and arrive at comparative resolutions as people. Profound learning accomplishes this with brain organizations, which are multifaceted forecasts. HRV is used to define the variation in time interval over two successive QRS waves. Deep Learning uses multi-layered algorithms, also called Neural Networks to teach the computers to work in patterns similar to the human brain. The prospective approach will be deployed to assist doctors in VTA detection and hence expedited, which is critical for effective treatment will be improved by preprocessing, integrating the data, that helps in better accuracy to predicate over different features extraction, so it is possible to increase prediction accuracy to 83.8%.