Title: Enhanced heart disease diagnosis via deep learning ECG signal analysis

Authors: Sunkara Venkata Kishore Babu; Mukkoti Maruthi Venkata Chalapathi

Addresses: School of Computer Science and Engineering, VIT-AP University, Amaravathi, 522237, Andhra Pradesh, India ' School of Computer Science and Engineering, VIT-AP University, Amaravathi, 522237, Andhra Pradesh, India

Abstract: Heart disease poses a substantial health risk, especially among older adults, and remains a leading global cause of death. Early detection is crucial, emphasising the importance of ECG in cardiovascular diagnosis. Traditionally, manual examination of paper-based ECGs is common but presents challenges. To tackle this, a novel DSEM architecture is proposed for heart disease diagnosis. This architecture integrates MWF for data pre-processing and ISLBT for improved shape feature extraction. The framework includes pre-processing, feature extraction, and classification phases. Initially, ECG signals undergo MWF pre-processing for clean representations. Then, comprehensive feature extraction, including ISLBT for shape features, is conducted. The DSEM architecture involves training bi-GRU, RNN, and LSTM models to capture various data aspects. The higher-level ImLinkNet model integrates predictions for final diagnosis. When TD = 90, the proposed method achieves a higher accuracy of 95% as compared to the other models.

Keywords: heart disease diagnosis; deep stacked ensemble architecture; MWF; ISLBT; ImLinkNet.

DOI: 10.1504/IJBRA.2026.152613

International Journal of Bioinformatics Research and Applications, 2026 Vol.22 No.2, pp.178 - 210

Received: 11 Jun 2024
Accepted: 02 Sep 2024
Published online: 31 Mar 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article