ABSTRACT
Cardiac disease is now a major cause of death for people affected by COVID-19. For the past five years, the death rate of people affected by the cardiac disease has increased a lot. In recent years, many deep learning models have provided prominent results for predicting it from different UCI heart disease data and other ECG data. Cardiac disease can be predicted from medical diagnosis and electrocardiogram data. Even though many types of detection for cardiac disease are available, ECG plays a major role in identifying it accurately. However, still, there is some gap in identifying the correct data, cleaning the unwanted features with popular methods, and optimizing it for better accuracy. In this paper, we propose a deep learning model, such as an Extreme Learning Machine (ELM), for predicting cardiac disease from the benchmark dataset, such as the MIT-BIH Arrhythmia dataset available in the PhysioNet database. The Principal Component Analysis is used to extract and identify the best features. Transfer learning is additionally used with kernel ELM for the improvement of the classification performance of ELM. Finally, the proposed Extreme Learning Machine model classifies cardiac disease with a promising result of 98.50% accuracy. In future research, it can be predicted in various datasets for performance improvement by selecting all other ensemble models. © 2022 Seventh Sense Research Group.