(B-078) Stakeholder Analysis of Responses to the FDA’s Discussion Paper on Artificial Intelligence and Machine Learning in Drug and Biological Product Development

Background: Artificial intelligence (AI) and machine learning (ML) are increasingly used in drug development, offering potential to accelerate discovery, refine clinical trials, and enhance post-market safety monitoring. However, these technologies raise complex regulatory challenges concerning oversight, transparency, validation, and data protection, requiring updated frameworks to ensure their safe and ethical integration.

Objectives: This study explores stakeholder perspectives on the regulatory and operational challenges posed by AI/ML in pharmaceutical research. It focuses on key areas raised by the U.S. Food and Drug Administration (FDA): model transparency, bias mitigation, data integrity, validation strategies, and adaptive governance approaches. The study also considers stakeholder concerns around proprietary protections and cross-agency harmonization.

Methods: A qualitative content analysis was conducted on public comments submitted in response to the FDA’s 2023 discussion paper, “Using Artificial Intelligence and Machine Learning in the Development of Drug and Biological Products.” The dataset comprises 69 comments from 64 stakeholders representing industry, academia, regulatory bodies, patient groups, and technology developers. Comments were thematically coded using an iterative approach to identify major regulatory priorities, implementation barriers, and policy recommendations. Comparative analysis was performed across stakeholder categories to assess points of convergence and divergence.

Results: Stakeholders broadly supported risk-based oversight models tailored to the context of AI application. Industry respondents emphasized regulatory flexibility and intellectual property protections, while other groups highlighted the importance of explainability, accountability, and equitable implementation. Suggested governance mechanisms included regulatory sandboxes, post-market monitoring, and secure, privacy-preserving data infrastructures such as federated learning and blockchain systems.

Conclusions: Stakeholder feedback underscores the need for a harmonized, adaptive regulatory framework that supports innovation while safeguarding public trust. Integrating continuous validation, scalable transparency measures, and context-sensitive oversight will be critical to the responsible adoption of AI in drug development.