Synthetic Data Generation Using Generative AI to Support Biomedical Innovation: A Health Policy Perspective

Synthetic Data Generation Using Generative AI to Support Biomedical Innovation: A Health Policy Perspective first page

Policy Brief

Synthetic Data Generation Using Generative AI to Support Biomedical Innovation: A Health Policy Perspective

Published date

June 11, 2025

Executive Summary

Regulators and payers globally are exploring the potential of synthetic data as one of many applications of generative artificial intelligence (AI) to support both operations and decision-making in medical product development. This ongoing exploration has highlighted a current need to identify and develop practical considerations associated with synthetic data generation use in this context. In this policy brief, we explore these areas through a discussion of current synthetic data management tools and best practices, ethical considerations for the generation and application of synthetic data, and regulatory developments to date. We recommend specific steps that regulatory stakeholders and practitioners may take to develop and describe regulatory fit-for-use synthetic data. Lastly, we offer a risk-based credibility assessment framework that could be helpful to for those managing synthetic data derived from generative AI applications.

Read the full paper here.

Duke-Margolis Authors

Dr. Rachele Hendricks-Sturrup, Research Director, Real World Evidence

Rachele Hendricks-Sturrup, DHSc, MSc, MA

Research Director, Real-World Evidence
Senior Team Member

Nora Emmott Headshot

Nora Emmott, MPH

Policy Research Associate

Maryam Nafie 2024 Headshot

Maryam Nafie

Policy Analyst