Abstract
Validating federated learning frameworks on real clinical data is an essential step between proof-of-concept demonstrations in controlled synthetic environments and deployment in real multicenter healthcare settings. A prior architectural study by the same authors (Tertulino and Alencar, 2026) demonstrated, on a synthetic six-feature benchmark, that server-side adaptive optimization acts as a temporal denoiser for Differential Privacy noise, addressing an open challenge identified in the original pipeline work (Tertulino, 2025). This study used synthetically generated data and explicitly identified real-world validation as a priority future direction.
This work addresses this gap by validating the FedCVR framework on five publicly available real cardiovascular datasets (Framingham, Cleveland, Hungarian, Switzerland, and Long Beach VA), harmonized to the 13-attribute UCI Heart Disease schema and configured as a heterogeneous federated scenario with leave-one-institution-out cross-validation. Results demonstrate that FedCVR preserves its adaptive advantage on real data, achieving an F1-Score of 79.2% and AUC of 0.96 under the operational privacy budget (noise multiplier = 0.8, privacy budget epsilon approximately 4.2), while statistically outperforming standard FedAvg on all evaluated metrics (paired t-tests, all p < 0.05).
Conclusion
This study validates the effectiveness of applying adaptive federated aggregation on real-world cardiovascular data, providing important empirical support for privacy-preserving data sharing in healthcare.
Blogger's Review: This paper empirically validates the effectiveness of federated learning on real datasets, showcasing the advantages of adaptive optimization within a differential privacy framework. It opens new avenues for protecting data privacy in healthcare, and future research could explore optimizing the balance between privacy budgets and model performance.