A Robust Pipeline for Differentially Private Federated Learning on Imbalanced Clinical Data using SMOTETomek and FedProx

Rodrigo Tertulino

doi:10.5753/jbcs.2026.6361

Authors

Rodrigo Tertulino IFRN https://orcid.org/0000-0002-7594-9312

DOI:

https://doi.org/10.5753/jbcs.2026.6361

Keywords:

Federated Learning, Differential Privacy, Cardiovascular Disease Prediction, Privacy-Utility Trade-off, SMOTETomek, Machine Learning in Healthcare

Abstract

Federated Learning (FL) offers a groundbreaking approach to collaborative health research, enabling model training on decentralized data while safeguarding patient privacy. FL offers formal security guarantees when combined with Differential Privacy (DP). The integration of these technologies, however, introduces a significant trade-off between privacy and clinical utility, a challenge further complicated by the severe class imbalance often present in medical datasets. The research presented herein addresses these interconnected issues through a systematic, multi-stage analysis. An FL framework was implemented for cardiovascular risk prediction, where initial experiments showed that standard methods struggled with imbalanced data, resulting in a Recall of zero. To overcome such a limitation, we first integrated the hybrid Synthetic Minority Over-sampling Technique with Tomek Links (SMOTETomek) at the client level, successfully developing a clinically useful model. Subsequently, the framework was optimized for non-IID data using a tuned FedProx algorithm. Our final results reveal a clear, non-linear trade-off between the privacy budget (ε) and model utility. An optimal operational region was identified where formal differential privacy guarantees with a single-digit privacy budget (ε ≈ 9.0 ) can be achieved while maintaining high clinical utility (Recall ≈ 77% ) and discriminative power (ROC-AUC ≈ 0.82 ). Although single-digit values ε are more relaxed than the tight budgets studied in theoretical DP literature, they represent a pragmatic operating point commonly reported in applied healthcare FL research. Ultimately, our study provides a practical methodological blueprint for creating effective, secure, and accurate diagnostic tools that can be applied to real-world, heterogeneous healthcare data.

Downloads

Download data is not yet available.

Author Biography

Rodrigo Tertulino, IFRN

Doutor em Engenharia Informática pela Universidade de Coimbra, Portugal. Possui Mestrado em Ciência da Computação pela UERN/UFERSA, graduação em Sistemas de Informação e MBA em Gestão de Negócios. Atualmente, é Professor de Redes de Computadores no IFRN, com dedicação exclusiva. Suas áreas de atuação incluem Redes e Sistemas Distribuídos, com ênfase na avaliação de desempenho de sistemas em rede e no gerenciamento de redes. Em Engenharia de Software, dedica-se ao estudo de métodos ágeis e sua integração com abordagens tradicionais, bem como ao desenvolvimento de software orientado a objetos, com foco em refatorações e frameworks. Na área de Segurança da Informação, atua com segurança em aplicações Web e Testes de Invasão (PenTest). Desenvolve ainda pesquisas relacionadas à privacidade e segurança em sistemas de saúde eletrônicos (EHR), com ênfase na aplicação de tecnologias Blockchain para proteção de dados e controle de acesso.

References

Abhishek, Bhagat, H. V., and Singh, M. (2023). A machine learning model for the early prediction of cardiovascular disease in patients. In 2023 Second International Conference on Advances in Computational Intelligence and Communication (ICACIC), pages 1-5. DOI: 10.1109/ICACIC59454.2023.10435210.

Ahmed, S., Kaiser, M. S., Chaki, S., Aloteibi, S., and Moni, M. A. (2025). Federated learning model with dynamic scoring-based client selection for diabetes diagnosis. Knowledge-Based Systems, 320:113662. DOI: 10.1016/j.knosys.2025.113662.

Ali, A., Pasha, M. F., Guerrieri, A., Guzzo, A., Sun, X., Saeed, A., Hussain, A., and Fortino, G. (2023). A novel homomorphic encryption and consortium blockchain-based hybrid deep learning model for industrial internet of medical things. IEEE Transactions on Network Science and Engineering, 10(5):2402-2418. DOI: 10.1109/TNSE.2023.3285070.

Dubey, M., Tembhurne, J., and Makhijani, R. (2025). Enhancing federated learning through differential privacy: Introducing FedHybrid for multicenter diverse heart disease datasets. IEEE Trans. Emerg. Top. Comput. Intell., pages 1-14. DOI: 10.1109/TETCI.2025.3575420.

Fedesoriano (2021). Stroke prediction dataset. Available at: [link].

Gaber, A., Abdeltwab, H., and Elbatt, T. (2024). Fedcvd: Towards a scalable, privacy-preserving federated learning model for cardiovascular diseases prediction. In Proceedings of the 2024 8th International Conference on Machine Learning and Soft Computing, ICMLSC '24, page 7–11, New York, NY, USA. Association for Computing Machinery. DOI: 10.1145/3647750.3647752.

Gupta, A., Misra, S., Pathak, N., and Das, D. (2023). Fedcare: Federated learning for resource-constrained healthcare devices in iomt system. IEEE Transactions on Computational Social Systems, 10(4):1587-1596. DOI: 10.1109/TCSS.2022.3232192.

Hudaib, A., Obeid, N., Albashayreh, A., Mosleh, H., Tashtoush, Y., and Hristov, G. (2025). Exploring the implementation of federated learning in healthcare: a comprehensive review. Cluster Comput., 28(5). DOI: 10.1007/s10586-024-05014-0.

Ismail, A. R., Azhary, M. Z. R., and Hitam, N. A. (2025). Evaluating adan vs. adam: An analysis of optimizer performance in deep learning. Communications in Computer and Information Science, 2381 CCIS:251 – 263. DOI: 10.1007/978-3-031-82931-4_19.

Jensen, B. K., Cline, M., and Guynes, C. S. (2007). Hippa, privacy and organizational change: a challenge for management. SIGCAS Comput. Soc., 37(1):12–17. DOI: 10.1145/1273353.1273354.

Khimani, A., Hornback, A., Jain, N., Avula, P., Jaishankar, A., and Wang, M. D. (2024). Predicting cardiovascular disease risk in tobacco users using machine learning algorithms. In 2024 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pages 1-5. DOI: 10.1109/EMBC53108.2024.10782885.

Khleel, N. A. A. and Nehéz, K. (2023). A novel approach for software defect prediction using CNN and GRU based on SMOTE tomek method. J. Intell. Inf. Syst., 60(3):673-707. DOI: 10.1007/s10844-023-00793-1.

Li, T., Sahu, A. K., Zaheer, M., Sanjabi, M., Talwalkar, A., and Smith, V. (2020). Federated optimization in heterogeneous networks. In Proceedings of the 3rd Conference on Machine Learning and Systems (MLSys), volume 2, pages 429-450. DOI: 10.48550/arxiv.1812.06127.

Liu, J., Wang, J. H., Rong, C., Xu, Y., Yu, T., and Wang, J. (2021). FedPA: An adaptively partial model aggregation strategy in federated learning. Comput. Netw., 199(108468):108468. DOI: 10.1016/j.comnet.2021.108468.

Liu, Y., Huang, A., Luo, Y., Huang, H., Liu, Y., Chen, Y., Feng, L., Chen, T., Yu, H., and Yang, Q. (2020). Fedvision: An online visual object detection platform powered by federated learning. Proceedings of the AAAI Conference on Artificial Intelligence, 34:13172-13179. DOI: 10.1609/aaai.v34i08.7021.

Lopez, E. O., Ballard, B. D., and Jan, A. (2023). Cardiovascular disease. StatPearls Publishing. DOI: 10.1378/chest.112.3_supplement.64s.

Lu, J. (2025). Ct image data analysis framework based on computer deep learning: Cardiovascular disease prediction. In Proceedings of the 2025 International Conference on Health Big Data, HBD '25, page 23–27, New York, NY, USA. Association for Computing Machinery. DOI: 10.1145/3733006.3733010.

Madathil, N. T., Dankar, F. K., Gergely, M., Belkacem, A. N., and Alrabaee, S. (2025). Revolutionizing healthcare data analytics with federated learning: A comprehensive survey of applications, systems, and future directions. Comput. Struct. Biotechnol. J., 28:217-238. DOI: 10.1016/j.csbj.2025.06.009.

McMahan, B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B. A. (2017). Communication-efficient learning of deep networks from decentralized data. In Artificial intelligence and statistics, pages 1273-1282. PMLR. DOI: 10.48550/arxiv.1602.05629.

Mensah, G. A., Fuster, V., Murray, C. J. L., Roth, G. A., and Global Burden of Cardiovascular Diseases and Risks Collaborators (2023). Global burden of cardiovascular diseases and risks, 1990-2022. J. Am. Coll. Cardiol., 82(25):2350-2473. DOI: 10.1016/j.jacc.2023.11.007.

Mittal, S., Gupta, S., Bansal, K., and Aggarwal, G. (2024). Democratizing gdpr compliance: Ai-driven privacy policy interpretation. In Proceedings of the 2024 Sixteenth International Conference on Contemporary Computing, IC3-2024, page 735–743, New York, NY, USA. Association for Computing Machinery. DOI: 10.1145/3675888.3676142.

Naresh, V. S. and Varma, G. T. (2025). Privacy-enhanced heart stroke detection using federated learning and homomorphic encryption. Smart Health, 37(100594):100594. DOI: 10.1016/j.smhl.2025.100594.

Naseri, M., Fernandez-Marques, J., Gao, Y., and Pan, H. (2024). Privacy-preserving federated learning using flower framework. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, KDD '24, page 6422–6423, New York, NY, USA. Association for Computing Machinery. DOI: 10.1145/3637528.3671447.

Nguyen, D.-K., Li, A.-H. A., Lai, Y.-J., Chiu, Y.-L., Phan, D.-V., Chien, T.-Y., and Chan, C.-L. (2025). Enhanced performance and data privacy in lung nodule classification via federated deep learning approach. In Proceedings of the 2024 7th International Conference on Healthcare Service Management, ICHSM '24, page 11–16, New York, NY, USA. Association for Computing Machinery. DOI: 10.1145/3704239.3704247.

Nishio, T. and Yonetani, R. (2019). Client selection for federated learning with heterogeneous resources in mobile edge. In ICC 2019 - 2019 IEEE International Conference on Communications (ICC), pages 1-7. DOI: 10.1109/ICC.2019.8761315.

Paul, J., Annamalai, M. S. M. S., Ming, W., Badawi, A. A., Veeravalli, B., and Aung, K. M. M. (2021). Privacy-preserving collective learning with homomorphic encryption. IEEE Access, 9:132084-132096. DOI: 10.1109/ACCESS.2021.3114581.

Reddi, S., Charles, Z., Zaheer, M., Garrett, Z., Rush, K., Konečný, J., Kumar, S., and McMahan, H. B. (2021). Adaptive federated optimization. DOI: 10.48550/arxiv.2003.00295.

Reisizadeh, A., Tziotis, I., Hassani, H., Mokhtari, A., and Pedarsani, R. (2022). Straggler-resilient federated learning: Leveraging the interplay between statistical accuracy and system heterogeneity. IEEE J. Sel. Areas Inf. Theory, 3(2):197-205. DOI: 10.1109/JSAIT.2022.3205475.

Saxena, K., Sharma, R., et al. (2016). Efficient heart disease prediction system. Procedia Computer Science, 85:962-969. DOI: 10.1016/j.procs.2016.05.288.

Silva, R. (2025). A comparative benchmark of federated learning strategies for mortality prediction on heterogeneous and imbalanced clinical data. Available at:[link].

Suriyakumar, V. M., Papernot, N., Goldenberg, A., and Ghassemi, M. (2021). Chasing your long tails: Differentially private prediction in health care settings. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, FAccT '21, page 723–734, New York, NY, USA. Association for Computing Machinery. DOI: 10.1145/3442188.3445934.

Tertulino, R. and Alencar, L. (2026). A robust framework for secure cardiovascular risk prediction: An architectural case study of differentially private federated learning. Peer-to-Peer Networking and Applications, 19(4):105. DOI: 10.1007/s12083-026-02265-z.

Thumula, K., Holla, H., Gutti, C., Sasikumar, A. A., and Gogineni, H. (2025). Privfed: Protecting user privacy in federated learning systems through differential privacy. In 2025 8th International Conference on Electronics, Materials Engineering & Nano-Technology (IEMENTech), pages 1-6. DOI: 10.1109/IEMENTech65115.2025.10959665.

Wang, X., Liesaputra, V., Liu, Z., Wang, Y., and Huang, Z. (2024). An in-depth survey on deep learning-based motor imagery electroencephalogram (EEG) classification. Artif. Intell. Med., 147:102738. DOI: 10.1016/j.artmed.2023.102738.

Yurdem, B., Kuzlu, M., Gullu, M. K., Catak, F. O., and Tabassum, M. (2024). Federated learning: Overview, strategies, applications, tools and future directions. Heliyon, 10(19):e38137. DOI: 10.1016/j.heliyon.2024.e38137.

Zhang, Y., Chen, G., Xu, Z., Wang, J., Zeng, D., Li, J., Wang, J., Qi, Y., and King, I. (2024). Fedcvd: The first real-world federated learning benchmark on cardiovascular disease data. arXiv preprint arXiv:2411.07050. DOI: 10.48550/arxiv.2411.07050.