Privacy-Utility Trade-offs in Neural Networks for Medical Population Graphs: Insights from Differential Privacy and Graph Structure

TL;DR

This paper explores the balance between privacy and utility in neural networks applied to medical population graphs, analyzing differential privacy impacts and graph structure effects on model performance.

Contribution

It provides an empirical study of differentially private graph neural networks in the medical domain, highlighting privacy-utility trade-offs and the influence of graph structure.

Findings

Privacy-utility trade-offs vary with privacy levels.

Graph homophily correlates with model accuracy.

Membership inference attacks reveal privacy vulnerabilities.

Abstract

We initiate an empirical investigation into differentially private graph neural networks on population graphs from the medical domain by examining privacy-utility trade-offs at different privacy levels on both real-world and synthetic datasets and performing auditing through membership inference attacks. Our findings highlight the potential and the challenges of this specific DP application area. Moreover, we find evidence that the underlying graph structure constitutes a potential factor for larger performance gaps by showing a correlation between the degree of graph homophily and the accuracy of the trained model.