An Enhanced Privacy-preserving Federated Few-shot Learning Framework for Respiratory Disease Diagnosis

2025-07-10

AI Summary

To address the dual challenges of scarce labeled data and stringent patient privacy requirements in respiratory disease diagnosis, this paper proposes a privacy-enhanced federated few-shot learning framework. The method integrates meta-stochastic gradient descent to mitigate overfitting in few-shot settings, incorporates Gaussian differential privacy to inject calibrated noise into uploaded gradients—thereby preventing medical image reconstruction—and employs weighted averaging for model aggregation to accommodate heterogeneous data distributions across multi-institutional collaborators. Experiments demonstrate that the framework significantly improves diagnostic accuracy under rigorous privacy guarantees (ε ≤ 2.0, δ = 10⁻⁵) across cross-institutional, cross-class, and cross-device scenarios. To the best of our knowledge, this is the first work to simultaneously achieve strong privacy preservation, robust generalization, and clinical practicality for few-shot federated learning in respiratory imaging diagnosis.

Technology Category

• Machine Learning: Privacy
• Computer Vision: Bias, Fairness & Privacy
• Search and Optimization: Learning to Search

Application Category

• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Security and Privacy: Privacy-enhancing technologies
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

The labor-intensive nature of medical data annotation presents a significant challenge for respiratory disease diagnosis, resulting in a scarcity of high-quality labeled datasets in resource-constrained settings. Moreover, patient privacy concerns complicate the direct sharing of local medical data across institutions, and existing centralized data-driven approaches, which rely on amounts of available data, often compromise data privacy. This study proposes a federated few-shot learning framework with privacy-preserving mechanisms to address the issues of limited labeled data and privacy protection in diagnosing respiratory diseases. In particular, a meta-stochastic gradient descent algorithm is proposed to mitigate the overfitting problem that arises from insufficient data when employing traditional gradient descent methods for neural network training. Furthermore, to ensure data privacy against gradient leakage, differential privacy noise from a standard Gaussian distribution is integrated into the gradients during the training of private models with local data, thereby preventing the reconstruction of medical images. Given the impracticality of centralizing respiratory disease data dispersed across various medical institutions, a weighted average algorithm is employed to aggregate local diagnostic models from different clients, enhancing the adaptability of a model across diverse scenarios. Experimental results show that the proposed method yields compelling results with the implementation of differential privacy, while effectively diagnosing respiratory diseases using data from different structures, categories, and distributions.

Problem

• Addresses limited labeled data for respiratory disease diagnosis
• Ensures privacy protection in federated medical data sharing
• Mitigates overfitting in neural networks with few-shot learning

Innovation

• Federated few-shot learning for respiratory disease diagnosis
• Meta-stochastic gradient descent to prevent overfitting
• Differential privacy noise to protect gradient leakage