Comparative Analysis of Machine Learning Algorithms for Early Heart Disease Detection Using ECG Data

Abstract

Cardiovascular Diseases (CVDs), is one of the leading causes of death worldwide,highlighting the importance of early identification for timely treatment. This studymarks a comparative analysis based on the Machine Learning (ML) approach targeted toward heart disease detection using the Electrocardiogram (ECG)as its input data. Methods such as Discrete Wavelet Transform (DWT) for feature extraction and Recursive Feature Elimination (RFE) for feature selection were applied to optimize model performance. The ML models evaluated included Vision Transformer (ViT), Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Multi-Layer Perceptron (MLP). These models were trained and tested on a dataset of 986 patients to assess their predictive accuracy. The results showed that MLP achieved the highest accuracy of 99.3%, followed by LSTM and CNN. These findings highlight the capability of ML to improve early detection of heart disease. Future research may include enhancing generalizability by including larger and more diverse datasets, hybrid models, and real-time diagnostic tools to further improve prediction accuracy and extend clinical applications.