Democratising Clinical AI through Dataset Condensation for Classical Clinical Models

2026-03-10

AI Summary

This work addresses the challenge of privacy-preserving medical data sharing by proposing the first differentially private dataset condensation framework tailored for non-differentiable clinical models—such as decision trees and Cox regression—which are widely used in practice but incompatible with existing gradient-based condensation methods. Leveraging zeroth-order optimization, the approach generates synthetic data using only model outputs, eliminating the need for gradient information. Evaluated across six datasets spanning classification and survival analysis tasks, the method demonstrates strong performance preservation under rigorous privacy guarantees. This study achieves, for the first time, model-agnostic, high-fidelity medical data condensation that is both privacy-safe and clinically deployable, thereby advancing data democratization within real-world clinical modeling ecosystems.

Technology Category

• Machine Learning: Privacy
• Computer Vision: Diffusion Models for Vision
• Data Mining & Knowledge Management: Data Compression

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Dataset condensation (DC) learns a compact synthetic dataset that enables models to match the performance of full-data training, prioritising utility over distributional fidelity. While typically explored for computational efficiency, DC also holds promise for healthcare data democratisation, especially when paired with differential privacy, allowing synthetic data to serve as a safe alternative to real records. However, existing DC methods rely on differentiable neural networks, limiting their compatibility with widely used clinical models such as decision trees and Cox regression. We address this gap using a differentially private, zero-order optimisation framework that extends DC to non-differentiable models using only function evaluations. Empirical results across six datasets, including both classification and survival tasks, show that the proposed method produces condensed datasets that preserve model utility while providing effective differential privacy guarantees - enabling model-agnostic data sharing for clinical prediction tasks without exposing sensitive patient information.

Problem

• dataset condensation
• clinical AI
• non-differentiable models
• differential privacy
• data democratisation

Innovation

• dataset condensation
• differential privacy
• zero-order optimization
• non-differentiable models
• clinical AI democratization