A Hybrid CNN-Random Forest Framework for Interpretable ECG Arrhythmia Detection

DOI

10.2991/978-94-6239-713-2_13How to use a DOI?

Keywords

ECG arrhythmia detection; convolutional neural network; random forest; hybrid deep learning; SHAP interpretability; cardiac classification; wearable health monitoring

Abstract

An automated clinical decision support based on cardiac arrhythmia detection from electrocardiogram (ECG) signals is still a burning issue, especially where high sensitivity, robustness under realistic signal conditions, and post-hoc interpretability are required. This paper proposes a Hybrid Convolutional Neural Network-Random Forest (CNN-RF) framework that addresses these challenges by employing a two-stage pipeline that uses an optimized 1-D CNN, which is trained by ECG-specific data augmentation that serves as a deep feature extractor and outputs only compact 64-dimensional morphological embeddings from raw ECG segments. Then the embeddings are classified by a regularized Random Forest ensemble. Shapley Additive Explanations (SHAP) is used for the Random Forest component to give clinically meaningful post-hoc feature attribution. The proposed Hybrid CNN-RF framework achieved 95.1% accuracy, 95.2% sensitivity, 94.9% specificity, a 95.1% F1-score, and an AUC-ROC of 0.950, matching the standalone CNN while providing interpretable features that a pure deep learning model alone cannot provide. The LSTM baseline showed critical performance degradation with the sensitivity and accuracy of 13.3% and 56.2%, respectively. These indicate that the proposed framework provides a reliable, robust, and interpretable alternative for automated arrhythmia detection, which can be improved and integrated for real-world wearable and clinical deployment scenarios.

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  - Muazu Abubakar Muhammad
AU  - Tarun Kumar Agrawal
PY  - 2026
DA  - 2026/06/25
TI  - A Hybrid CNN-Random Forest Framework for Interpretable ECG Arrhythmia Detection
BT  - Proceedings of the International Conference on Advances in Computing Technology and Artificial Intelligence (COMPUTATIA 2026)
PB  - Atlantis Press
SP  - 176
EP  - 191
SN  - 2589-4919
UR  - https://doi.org/10.2991/978-94-6239-713-2_13
DO  - 10.2991/978-94-6239-713-2_13
ID  - Muhammad2026
ER  -