Deployment-Oriented Quantization of an ECGWavePuzzle-Based Personalization Pipeline for Arrhythmia Classification

• Guilherme Silva UFOP
• Arthur Negrão UFOP
• Pedro Silva UFOP
• Eduardo Luz UFOP

DOI: https://doi.org/10.5753/sbcas.2026.21746

Resumo

This paper evaluates INT8 quantization and mixed-precision execution in an ECGWavePuzzle-based personalization pipeline for arrhythmia classification. The SSL framework follows prior work, and the contribution is a deployment-oriented analysis under reduced numerical precision. On MITBIH with ANSI/AAMI patient-wise evaluation, head int8 achieves 0.9013 ± 0.0198 Accuracy and 0.5364± 0.0416 Macro-F1 on H200. FP16 also improves Macro-F1 over BF16 on H200 (0.5252 vs. 0.4969) with slightly lower runtime (2920 s vs. 3003 s). However, Jetson Nano execution remains expensive, requiring about 60048 s. The results show that selective quantization is feasible, but online personalization remains the main bottleneck for embedded deployment.

Referências

An, X., Shi, S., Wang, Q., Yu, Y., and Liu, Q. (2024). Research on a lightweight arrhythmia classification model based on knowledge distillation for wearable single-lead ecg monitoring systems. Sensors, 24(24):7896.

ANSI/AAMI (2008). Testing and reporting performance results of cardiac rhythm and ST segment measurement algorithms. American National Standards Institute, Inc. (ANSI), Association for the Advancement of Medical Instrumentation (AAMI). ANSI/AAMI/ISO EC57, 1998-(R)2008.

De Chazal, P., O’Dwyer, M., and Reilly, R. B. (2004). Automatic classification of heartbeats using ECG morphology and heartbeat interval features. IEEE transactions on biomedical engineering, 51(7):1196–1206. Publisher: IEEE.

Goldberger, A. L., Amaral, L. A., Glass, L., Hausdorff, J. M., Ivanov, P. C., Mark, R. G., Mietus, J. E., Moody, G. B., Peng, C.-K., and Stanley, H. E. (2000). Physiobank, physiotoolkit, and physionet: components of a new research resource for complex physiologic signals. circulation, 101(23):e215–e220.

Gupta, U., Paluru, N., Nankani, D., Kulkarni, K., and Awasthi, N. (2024). A comprehensive review on efficient artificial intelligence models for classification of abnormal cardiac rhythms using electrocardiograms. Heliyon, 10(5):e26787.

Liu, H., Zhao, Z., and She, Q. (2021). Self-supervised ecg pre-training. Biomedical Signal Processing and Control, 70:103010.

Liu, W., Guo, Q., Chen, S., Chang, S., Wang, H., He, J., and Huang, Q. (2023). A fully-mapped and energy-efficient fpga accelerator for dual-function ai-based analysis of ecg. Frontiers in Physiology, 14:1079503.

Luz, E. J. d. S., Schwartz, W. R., Cámara-Chávez, G., and Menotti, D. (2016). ECG-based heartbeat classification for arrhythmia detection: A survey. Computer methods and programs in biomedicine, 127:144–164. Publisher: Elsevier.

Mehari, T. and Strodthoff, N. (2022). Self-supervised representation learning from 12-lead ecg data. Computers in Biology and Medicine, 141:105114.

Phan, T., Le, D., Brijesh, P., Adjeroh, D., Wu, J., Jensen, M. O., and Le, N. (2022). Multimodality multi-lead ecg arrhythmia classification using self-supervised learning. In 2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI), pages 01–04. IEEE.

Silva, G., Silva, P., Moreira, G., Freitas, V., Gertrudes, J., and Luz, E. (2025). A systematic review of ecg arrhythmia classification: Adherence to standards, fair evaluation, and embedded feasibility. arXiv preprint arXiv:2503.07276.

Silva, G., Silva, P., Moreira, G., and Luz, E. (2024). Bridging the gap in ecg classification: Integrating self-supervised learning with human-in-the-loop amid medical equipment hardware constraints. In International Symposium on Applied Reconfigurable Computing, pages 63–74. Springer.

Silva, G. A. L. (2023). Self-supervised learning for arrhythmia classification. Thesis (master in computer science), Federal University of Ouro Preto, Ouro Preto.