Demography-Aware Personalized Federated Learning for Fair, Private, and Efficient Clinical Risk Prediction

📝 Abstract

We propose a demography-aware personalized federated learning framework that achieves fair, private, and efficient clinical risk prediction. Existing federated learning approaches for clinical applications often overlook demographic heterogeneity across sites, leading to biased predictions for underrepresented subgroups. By explicitly incorporating demographic information into the federated optimization process, our framework learns personalized models that are both locally adapted and globally coordinated. Differential privacy mechanisms ensure patient data remains protected throughout training, while an efficient communication protocol reduces the overhead of federated rounds.

📋 BibTeX Citation

@article{zhao2026demofl,
  title     = {Demography-Aware Personalized Federated Learning 
               for Fair, Private, and Efficient Clinical Risk 
               Prediction},
  author    = {Zhao, Puyang and Yue, Z. and Mi, N. and Zhang, H.},
  journal   = {Biomedical Signal Processing and Control},
  volume    = {114},
  pages     = {109312},
  year      = {2026},
  publisher = {Elsevier}
}

🔗 Related Publications