FedDP-PALD: A Privacy-Preserving Federated Latent Diffusion Framework with Prototype Aggregation for Medical Data Synthesis

2026-07-13

AI Summary

This work addresses the challenge of balancing privacy preservation and synthetic data utility in multi-institutional collaborative training by proposing a differentially private federated latent diffusion framework. The method integrates a gated multi-head attention mechanism with modality availability masks to uniformly handle heterogeneous medical signals such as chest X-rays and electrocardiograms. It introduces, for the first time, a Differentially Private Prototype Mixture Aggregation (DP-PMA) strategy that clips and perturbs class-level latent prototypes at the server. Under a strict privacy budget (ε = 1–8), the framework reduces membership inference attack success to an AUROC of 0.5016–0.5093, while models trained on the synthesized data achieve an F1 score of 0.8993 ± 0.0006 and AUROC of 0.9057 ± 0.0503—nearly matching real-data performance—and substantially enhances multimodal synthetic quality and clinical applicability under strong privacy guarantees.

Abstract

Medical images and physiological signals provide valuable information for accurate diagnosis. Developing diagnostic models often requires patient data from multiple institutions, although strict privacy regulations limit the sharing of sensitive clinical records. Federated learning enables multiple hospitals to train a shared model without exchanging raw data. However, existing methods face two problems: the information exchanged during training can reveal whether a patient's data were used, and synthetic data meant to replace real records often fail to preserve their predictive structure, which limits clinical use. To address this issue, we propose FedDP-PALD, a privacy-preserving federated latent diffusion framework for multimodal medical data synthesis under formal privacy guarantees. It jointly processes chest X-ray images and electrocardiogram (ECG) signals through gated multi-head attention with modality-availability masks, remaining effective even when a modality is missing. We also introduce Differentially Private Prototype Mixture Aggregation (DP-PMA), which clips class-level latent prototypes and adds calibrated Gaussian noise before combining them on the server to maintain $(ε, δ)$ differential privacy. We evaluate FedDP-PALD on PneumoniaMNIST, ChestMNIST, and MIT-BIH datasets, where differential privacy reduced summary-level attack AUROC from 0.6229 $\pm$ 0.0026 to between 0.5016 and 0.5093 for privacy budgets from $ε= 1$ to $ε= 8$. On the test data, synthetic-latent training achieved an F1 score of 0.8993 $\pm$ 0.0006 and an AUROC of 0.9057 $\pm$ 0.0503, close to the 0.9747 $\pm$ 0.0132 real-latent training. These results show that FedDP-PALD generates private synthetic representations that preserve useful decision performance while strongly resisting membership inference.

Problem

• federated learning
• medical data synthesis
• privacy preservation
• differential privacy
• synthetic data

Innovation

• Federated Learning
• Differential Privacy
• Latent Diffusion
• Prototype Aggregation
• Medical Data Synthesis