EU and US launch joint AI principles for medicines industry

The European Medicines Agency and the American Food and Drug Administration have jointly identified 10 principles which they categorise as representing good practice for AI usage in medicine processes, covering areas such as evidence monitoring and risk assessment.

The 10 principles laid down hold that AI use should be human-centric by design, take a risk-aware based approach, adhere to defined standards, be clear in the context of their usage, and require multidisciplinary expertise in overseeing the technology and its implementation.

It also requires clear data governance and documentation, fit-for-use model design and deployment, clear life-cycle management and using plain language for providing information about the usage for appropriate audiences

Nils Rauer, an AI expert with Pinsent Masons in Frankfurt, said that while the principles offered high level guidance, more detailed approaches would still be required for practical implementation.

“The overall concept is very much in line with the approach taken by the European legislator,” he explained.

“We see similarities in thinking when looking at the EU AI Act and the set of principles now published. For instance, both follow a risk-based approach, both appreciate the importance of adequate data governance and AI literacy and both highlight the need for decent and thought through model design and development.

“One principle is crucial – that AI technologies have a well-defined context of use. Defining in adequate detail what AI will actually do within a wider context is highly important. The regulatory obligations may very well differ and vary depending on what the AI does and what role may be attributed to the company."

The move by the two agencies comes as AI adoption across the medicines and life sciences industries continues to increase, with new legislation around pharmaceutical industries allowing for the broader use of AI around development and regulatory decision-making.

They also come in the wake of a bilateral meeting between EU health chiefs and the FDA in April 2024, as the EMA looks to establish AI rules for the medicines industry in like with the content’s guidance and legal requirements.

Catherine Drew, a life sciences specialist with Pinsent Masons, said the joint natucre of the guidance indicated the agencies recognised the need for international cooperation and alignment on AI engagement.

“It’s really important for ensuring that medicine development and approval keeps pace with broader developments in technology, with AI being a key one of them, to make sure innovative treatments are available to patients as soon as possible,” she said.

“To do that while ensuring that standards of safety and efficacy are not compromised through the use of AI is essential, which is recognised in the note and is supported by the proposed risk based approach.”