Learnable Wavelet-Based Deep Learning for ECG Arrhythmia Detection

DOI

10.2991/978-94-6239-789-7_50How to use a DOI?

Keywords

Electrocardiographic Signals; Discrete Wavelet Transform; Convolutional Neural Network; Bidirectional Long Short-Term Memory

Abstract

Accurate and automated detection of cardiac arrhythmias Using electrocardiogram (ECG) data is a significant research challenge. Cardiac arrhythmias have been a major cause of cardiovascular mortality. The work proposes a unique end-to-end deep learning (DL) system for automated arrhythmia classification on the MIT-BIH Arrhythmia Database. The proposed model uses a Learnable Discrete Wavelet Transform (LDWT) layer whose filter coefficients are jointly optimized with the downstream classifier via backpropagation, in contrast to previous approaches that rely on fixed-basis Continuous Wavelet Transform (CWT) scalograms as precomputed, non-trainable features. A hybrid classifier consisting of a one-dimensional Convolutional Neural Network (1D-CNN) for local morphological feature extraction, a Bidirectional Long Short-Term Memory (BiLSTM) network for temporal rhythm modeling, and a self-attention layer to weigh discriminative time steps is fed multi-scale sub-band coefficients generated by the LDWT, concatenated with FFT magnitude features. The AAMI EC57 inter-patient protocol is used to train and assess the entire pipeline, which divides beats into 5 standard categories: Normal (N), Supraventricular ectopic (S), Ventricular ectopic (V), Fusion (F), and Unknown/paced (Q). The novelty of the proposed work lies in the model’s ability to independently identify arrhythmia-discriminative frequency bands due to the differentiable wavelet foundation.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Sumita Lamba
AU  - Satender Kumar
AU  - Manoj Diwakar
PY  - 2026
DA  - 2026/10/08
TI  - Learnable Wavelet-Based Deep Learning for ECG Arrhythmia Detection
BT  - Proceedings of theInternational Conference on Innovation and Practices in Artificial Intelligence (ICIPAI 2026)
PB  - Atlantis Press
SP  - 747
EP  - 764
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-789-7_50
DO  - 10.2991/978-94-6239-789-7_50
ID  - Lamba2026
ER  -