Decoding the Heartbeat: A Computational Intelligence Framework Coupling Nonlinear ODE Dynamics and Deep Attention Networks for Cardiac Arrhythmia Detection
DOI:
https://doi.org/10.54361/LJMR.20.3.97Keywords:
Electrocardiogram (ECG), Inter-patient Splitting, Patient-wise Evaluation, Non-linear ODE Model, Limit Cycle Attractor, Discrete Wavelet Transform (DWT), 1D-CNN-BiGRU-Attention, McNemar Test, Ablation Study, AAMIAbstract
Background: Automated electrocardiogram (ECG) heartbeat classification plays a paramount role in the early diagnosis of cardiovascular diseases [1, 2, 10]. However, existing deep-learning approaches frequently suffer from severe data leakage caused by beat-level data splitting [6, 9], class imbalance in benchmark databases like MIT-BIH [13, 17, 25], and a lack of statistical significance testing and explicit ablation analyses validating complex hybrid network extensions.
Methodology: To overcome these methodological pitfalls, this paper presents a fully revised, patient-wise rigorous computational intelligence framework. First, a three-dimensional coupled non-linear Ordinary Differential Equation (ODE) limit-cycle model (McSharry-type) [19] generates synthetic cardiac signals to augment minority classes exclusively within the training partition. Second, raw ECG signals are processed using a 4th-order zero-phase Butterworth band-pass filter (0.5–45 Hz) [3] and a 60 Hz IIR notch filter [4]. Third, an inter-patient (patient-wise) 70/15/15 train/validation/test split protocol is strictly applied to the MIT-BIH Arrhythmia Database [25, 26] to eliminate data leakage. Isolated 360-sample heartbeat segments centered on R-peaks detected via Pan-Tompkins [21] or Elgendi [22] methods are classified into ANSI/AAMI EC57 categories (Normal, Supraventricular Ectopic Beats, Ventricular Ectopic Beats) [23] using a hybrid 1D-CNN-BiGRU network [9, 12, 14, 16] equipped with Bahdanau-type additive attention [18, 24]. Hand-crafted wavelet features extracted via Discrete Wavelet Transform (Symlet-4) [15, 20] serve as inputs for classical Random Forest and SVM baselines [5].Results: Under strict patient-wise evaluation, the proposed ODE-augmented 1D-CNN-BiGRU-Attention network achieved an overall test accuracy of 97.40%, with per-class F1-scores of 0.981 (N), 0.884 (SVEB), and 0.942 (VEB). McNemar’s exact test confirmed that the performance gain over the standalone 1D-CNN baseline [11, 12] was statistically significant (p < 0.001, beat-level and exploratory; to be re-verified). Systematic ablation experiments verified that ODE-based synthetic augmentation contributed +2.15% directly to SVEB sensitivity. Conclusion: Enforcing methodological rigor—specifically inter-patient partitioning [6], statistical significance testing, and ablation studies—is crucial for realistic clinical evaluation [1, 2]. Coupling analytical ODE dynamics [19] with an attention deep architecture [18, 24] gives promising patient-wise performance on MIT-BIH; validation on larger independent cohorts is required before any claim of clinical transferability.
Downloads
References
1. Luz, E. J. S., Schwartz, W. R., Cámara-Chávez, G., & Menotti, D. (2016). ECG-based heartbeat classification for arrhythmia detection: A survey. Computer Methods and Programs in Biomedicine, 127, 144–164. https://doi.org/10.1016/j.cmpb.2015.12.008
2. Hong, S., Zhou, Y., Shang, J., Xiao, C., & Sun, J. (2020). Opportunities and challenges of deep learning methods for electrocardiogram data: A systematic review. Computers in Biology and Medicine, 122, 103801. https://doi.org/10.1016/j.compbiomed.2020.103801
3. Clifford, G. D., Azuaje, F., & McSharry, P. E. (Eds.). (2006). Advanced Methods and Tools for ECG Data Analysis. Artech House. ISBN 9781580539661. https://books.google.com/books?vid=ISBN9781580539661
4. Sayadi, O., & Shamsollahi, M. B. (2008). ECG denoising and compression using a modified extended Kalman filter structure. IEEE Transactions on Biomedical Engineering, 55(9), 2240–2248. https://doi.org/10.1109/TBME.2008.921150
5. Ye, C., Kumar, B. V. K. V., & Coimbra, M. T. (2012). Heartbeat classification using morphological and dynamic features of ECG signals. IEEE Transactions on Biomedical Engineering, 59(10), 2930–2941. https://doi.org/10.1109/TBME.2012.2212896
6. De Chazal, P., O’Dwyer, M., & Reilly, R. B. (2004). Automatic classification of heartbeats using ECG morphology and heartbeat interval features. IEEE Transactions on Biomedical Engineering, 51(7), 1196–1206. https://doi.org/10.1109/TBME.2004.827359
7. Kachuee, M., Fazeli, S., & Sarrafzadeh, M. (2018). ECG heartbeat classification: A deep transferable representation. In Proc. IEEE International Conference on Healthcare Informatics (ICHI). arXiv:1805.00794. https://doi.org/10.48550/arXiv.1805.00794
8. Al Rahhal, M. M., Bazi, Y., AlHichri, H., Alajlan, N., Melgani, F., & Yager, R. R. (2016). Deep learning approach for active classification of electrocardiogram signals. Information Sciences, 345, 340–354. https://doi.org/10.1016/j.ins.2016.01.082
9. Kiranyaz, S., Ince, T., & Gabbouj, M. (2016). Real-time patient-specific ECG classification by 1-D convolutional neural networks. IEEE Transactions on Biomedical Engineering, 63(3), 664–675. https://doi.org/10.1109/TBME.2015.2468589
10. Hannun, A. Y., Rajpurkar, P., Haghpanahi, M., Tison, G. H., Bourn, C., Turakhia, M. P., & Ng, A. Y. (2019). Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network. Nature Medicine, 25(1), 65–69. https://doi.org/10.1038/s41591-018-0268-3
11. Acharya, U. R., Oh, S. L., Hagiwara, Y., Tan, J. H., Adam, M., Gertych, A., & San Tan, R. (2017). A deep convolutional neural network model to classify heartbeats. Computers in Biology and Medicine, 89, 389–396. https://doi.org/10.1016/j.compbiomed.2017.08.022
12. Acharya, U. R., Fujita, H., Oh, S. L., Hagiwara, Y., Tan, J. H., & Adam, M. (2017). Automated detection of arrhythmias using different intervals of tachycardia ECG segments with convolutional neural network. Information Sciences, 405, 81–90. https://doi.org/10.1016/j.ins.2017.04.012
13. Sellami, A., & Hwang, H. (2019). A robust deep convolutional neural network with batch-weighted loss for heartbeat classification. Expert Systems with Applications, 122, 75–84. https://doi.org/10.1016/j.eswa.2018.12.037
14. Oh, S. L., Ng, E. Y. K., Tan, R. S., & Acharya, U. R. (2018). Automated diagnosis of arrhythmia using combination of CNN and LSTM techniques with variable length heart beats. Computers in Biology and Medicine, 102, 278–287. https://doi.org/10.1016/j.compbiomed.2018.06.002
15. Wang, T., Lu, C., Sun, Y., Yang, M., Liu, C., & Ou, C. (2021). Automatic ECG classification using continuous wavelet transform and convolutional neural network. Entropy, 23(1), 119. https://doi.org/10.3390/e23010119
16. Yildirim, Ö. (2018). A novel wavelet sequence based on deep bidirectional LSTM network model for ECG signal classification. Computers in Biology and Medicine, 96, 189–202. https://doi.org/10.1016/j.compbiomed.2018.03.016
17. Rajesh, K. N. V. P. S., & Dhuli, R. (2018). Classification of imbalanced ECG beats using re-sampling techniques and AdaBoost ensemble classifier. Biomedical Signal Processing and Control, 41, 242–254. https://doi.org/10.1016/j.bspc.2017.12.004
18. Yao, Q., Wang, R., Fan, X., Liu, J., & Li, Y. (2020). Multi-class arrhythmia detection from 12-lead varied-length ECG using attention-based time-incremental convolutional neural network. Information Fusion, 53, 174–182. https://doi.org/10.1016/j.inffus.2019.06.024
19. McSharry, P. E., Clifford, G. D., Tarassenko, L., & Smith, L. A. (2003). A dynamical model for generating synthetic electrocardiogram signals. IEEE Transactions on Biomedical Engineering, 50(3), 289–294. https://doi.org/10.1109/TBME.2003.808805
20. Mallat, S. (2009). A Wavelet Tour of Signal Processing: The Sparse Way (3rd ed.). Academic Press. ISBN 9780123743701. https://books.google.com/books?vid=ISBN9780123743701
21. Pan, J., & Tompkins, W. J. (1985). A real-time QRS detection algorithm. IEEE Transactions on Biomedical Engineering, 32(3), 230–236. https://doi.org/10.1109/TBME.1985.325532
22. Elgendi, M. (2013). Fast QRS detection with an optimized knowledge-based method: Evaluation on 11 standard ECG databases. PLoS ONE, 8(9), e73557. https://doi.org/10.1371/journal.pone.0073557
23. Association for the Advancement of Medical Instrumentation. (2012). Testing and reporting performance results of cardiac rhythm and ST segment measurement algorithms (ANSI/AAMI EC57:2012). https://webstore.ansi.org/search/find?in=1&st=EC57
24. Bahdanau, D., Cho, K., & Bengio, Y. (2015). Neural machine translation by jointly learning to align and translate. In Proc. 3rd International Conference on Learning Representations (ICLR). arXiv:1409.0473. https://doi.org/10.48550/arXiv.1409.0473
25. Moody, G. B., & Mark, R. G. (2001). The impact of the MIT-BIH Arrhythmia Database. IEEE Engineering in Medicine and Biology Magazine, 20(3), 45–50. https://doi.org/10.1109/51.932724
26. Goldberger, A. L., Amaral, L. A. N., Glass, L., Hausdorff, J. M., Ivanov, P. C., Mark, R. G., Mietus, J. E., Moody, G. B., Peng, C.-K., & Stanley, H. E. (2000). PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation, 101(23), e215–e220. https://doi.org/10.1161/01.CIR.101.23.e215
27. Sun, J. (2023). Automatic cardiac arrhythmias classification using CNN and attention-based RNN network. Healthcare Technology Letters, 10(3), 53–61. https://doi.org/10.1049/htl2.12045
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Libyan Journal of Medical Research
This work is licensed under a Creative Commons Attribution 4.0 International License.
Open Access Policy
Libyan journal of medical Research (LJMR).is an open journal, therefore there are no fees required for downloading any publication from the journal website by authors, readers, and institution.
The journal applies the license of CC BY (a Creative Commons Attribution 4.0 International license). This license allows authors to keep ownership f the copyright of their papers. But this license permits any user to download , print out, extract, reuse, archive, and distribute the article, so long as appropriate credit is given to the authors and the source of the work.
The license ensures that the article will be available as widely as possible and that the article can be included in any scientific archive.
Editorial Policy
The publication of an article in a peer reviewed journal is an essential model for Libyan journal of medical Research (LJMR). It is necessary to agree upon standards of expected ethical behavior for all parties involved in the act of publishing: the author, the journal editorial, the peer reviewer and the publisher.
Any manuscript or substantial parts of it, submitted to the journal must not be under consideration by any other journal. In general, the manuscript should not have already been published in any journal or other citable form, although it may have been deposited on a preprint server. Authors are required to ensure that no material submitted as part of a manuscript infringes existing copyrights, or the rights of a third party.
Authorship Policy
The manuscript authorship should be limited to those who have made a significant contribution and intellectual input to the research submitted to the journal, including design, performance, interpretation of the reported study, and writing the manuscript. All those who have made significant contributions should be listed as co-authors.
Others who have participated in certain substantive aspects of the manuscript but without intellectual input should only be recognized in the acknowledgements section of the manuscript. Also, one of the authors should be selected as the corresponding author to communicate with the journal and approve the final version of the manuscript for publication in the LJMR.
Peer-review Policy
- All the manuscripts submitted to LJMR will be subjected to the double-blinded peer-review process;
- The manuscript will be reviewed by two suitable experts in the respective subject area.
- Reports of all the reviewers will be considered while deciding on acceptance/revision or rejection of a manuscript.
- Editor-In-Chief will make the final decision, based on the reviewer’s comments.
- Editor-In-Chief can ask one or more advisory board members for their suggestions upon a manuscript, before making the final decision.
- Associate editor and review editors provide administrative support to maintain the integrity of the peer-review process.
- In case, authors challenge the editor’s negative decision with suitable arguments, the manuscript can be sent to one more reviewer and the final decision will be made based upon his recommendations.
https://portal.issn.org/resource/ISSN/2413-6069
