AI is playing an increasingly important role in the development and regulation of medicines, offering new opportunities to improve efficiency, accuracy, and patient outcomes. In the pharmaceutical and biotechnology sectors, the use of AI has increased significantly in the past years. AI is used to generate and analyse evidence throughout the entire drug product lifecycle, from early nonclinical research and clinical trials to manufacturing processes and post-marketing safety monitoring.
To ensure that these technologies are used responsibly, the European Medicines Agency (EMA) and the U.S. Food and Drug Administration (FDA) have jointly established ten principles for good AI practice. These principles are designed to ensure that AI supports, rather than compromises, the fundamental regulatory requirements of quality, safety, and efficacy. As AI systems become more complex, careful management is required to maintain reliable performance and minimise risks to patients.
A key focus of the principles is a human-centric, ethical and risk-based approach, where the level of oversight and validation depends on the context in which an AI system is used. Strong data governance, transparent model development, and continuous monitoring in line with GxP requirements are essential to prevent errors, bias, or performance degradation over time. Clear communication about how AI systems function and their limitations is also critical for regulators, developers, and end users.
Through this collaboration, EMA and FDA aim to promote international harmonisation and support innovation in drug development while maintaining high standards of patient and animal safety. A complete overview of the ten principles is available on the EMA website.
What are your thoughts on the recent AI developments in relation to medicine development and registration? At Starodub we are following the progress of AI software with great interest. Feel free to reach out to us on [email protected] to discuss any possibilities or challenges you perceive for your company.
