Synthetic Data to Support Reproducible Clinical Research: Opportunities and Challenges

Enabling access of individual patient level data to interested researchers can be a powerful way to promote replicability and spark innovation through additional analyses. However, ensuring protection of patient privacy is a central concern in data sharing. In recent years, several research groups have proposed use of synthetic data generated from real data as a proxy to enable wider sharing of underlying data used for clinical research. In this panel, speakers will share their experiences with synthetic data generators. The panel covered two key aspects, 1) the ability of synthetic data to arrive at valid conclusions, and 2) the extent to which these methods can minimize the risk for re-identification of patients and features that make some synthetic data more robust against re-identification.