Differential Privacy: Theory and Practice of Synthetic Healthcare Data – Parisa Movahedi (Department of Computing)

As data sharing and analytics become essential to medical research and healthcare services, maintaining patient privacy while enabling valuable analysis poses a critical challenge. Differentially private (DP) synthetic data has emerged as a potential solution for sharing sensitive individual-level data. DP generative models offer a promising approach for generating realistic synthetic data that aims to maintain the original data’s central statistical properties while ensuring privacy by limiting the risk of disclosing sensitive information about individuals.

DP enables healthcare organizations to share data and conduct analyses without compromising patient confidentiality. However, balancing data utility and privacy poses challenges and potential pitfalls. We will discuss strategies for managing this balance, such as setting appropriate privacy budgets and understanding statistical trade-offs, while highlighting common mistakes that can undermine data quality or privacy protections.

This presentation will cover the core principles of DP, its specific applications in healthcare, and how it enables compliance with privacy regulations and enhances trust in data-driven healthcare.