The pharmaceutical industry, while in the early stages of incorporating artificial intelligence (AI) in drug development, is using AI throughout the product lifecycle. Industry also supports the European Medicines Agency’s (EMA) reflection paper on AI, as it reflects current industry practices, asserted Nicole Mahoney, executive director of global regulatory policy for Novartis, at DIA Europe on 12 March.
On the regulatory front, in the absence of a formal regulatory framework for AI and machine learning (ML), pharmaceutical manufacturers should adhere to the principles governing model informed drug development (MIDD) because many if these same principles apply, said Gabriel Westman, head of artificial intelligence for the Swedish Medicines Products Agency.
The session focused on the emerging AI regulatory environment with some industry perspectives on current practices.
In the EU, a framework has been proposed under the EU AI Act, which was passed by the European Parliament earlier this week. EMA has also sought comments on an AI reflection paper released in July 2023 that provides insight into when AI/ML technology can be used to develop drug products. (RELATED: EMA details lifecycle approach to AI/ML drug development in new reflection paper, Regulatory Focus 20 July 2023)
Industry perspective
Mahoney, who was speaking on behalf of the European Federation of Pharmaceutical Industries and Associations (EfPIA), offered a perspective on the industry adoption of AI.
She said that “it is safe to say that in industry, we are in the early stages of AI but it is being used across the lifecycle of the product” and for different purposes.
Mahoney said that in the drug discovery and development phase, AI is being used for drug target identification and selection; in the preclinical areas, the tool is being used for pharmacokinetic (PK) and pharmacodynamic (PD) studies. These tools can be especially useful in replacing animal studies, which she said is ‘a goal that we all have in common.”
In the clinical research space, AI models are being used for patient recruitment, to select sites for clinical studies, and for capturing clinical endpoints for clinical trials.
In the manufacturing area, AI tools are being used to support advanced pharmaceutical manufacturing and postmarket safety surveillance studies.
Mahoney said industry supports EMA’s reflections paper in that it captures current industry practices of leveraging AI in different ways and these models are being used throughout the lifecycle of a product. “We are happy that it covered the entire life cyce of product development and offered some examples of how regulators will provide oversight for different uses.”
Industry also supports the paper’s risk-based approach in according a higher level of regulatory scrutiny of AI systems for products that that pose a high risk.
Same model used for MIDD
On the regulatory front, Westman said that manufacturers that leverage AI/ML should adhere to the regulatory principles of MIDD.
He said that “there is nothing new in model informed drug development and 95% percent of those principles can be applied to AI.” He added that AI models are “becoming larger and more powerful” than MIDD models “but the basic principles are the same.”
Under both tools, however “the marketing authorization holder and the applicant still have full responsibility for the information in the data sets. There are no excuses just because you are using AI.” He added that sponsors should “make sure you can trust your data.”
Westman said that EMA is currently reviewing comments on its reflection paper, which is expected to be finalized sometime this year.
The workplan of the EMA’s Committee for Medicinal Products for Human Use’s (CHMP) Methodology Working Party (MWP) also includes the development of specific guidance on the use of AI in clinical development and the use of AI in pharmacovigilance.
**To continue reading this article please go to RAPS . **