Regulatory Interactions for AI in Medicines

Applicants and developers are expected to perform a regulatory impact and risk analysis of all AI/ML applications and are recommended to seek regulatory interactions when no clearly applicable written guidance is available.

Regulatory impact and timing

The regulatory impact is directly related to the phase in the medicinal product lifecycle and the weight of evidence the data will have in the intended setting. In cases where impact on regulatory decision-making is high, interaction with regulators is always recommended.

Timing of interactions should be guided by the regulatory impact and risk associated with using AI-based models. In high-impact cases, interaction may be crucial already at the planning stage. If development or use of a medicinal product critically relies on information from an AI/ML medical device, or the information may be included in the Summary of Product Characteristics, early regulatory interaction is also recommended.

EMA engagement pathways

Early interaction on experimental technology is provided by the EMA Innovation Task Force (ITF). Scientific advice and qualification of novel methodologies in medicines development are provided by the Scientific Advice Working Party (SAWP) of CHMP and CVMP.

Qualification advice or opinion refers to novel methodologies applied to medicinal product development where the methodology to be qualified would ideally be medical device/software agnostic.

Documentation for regulatory interaction

Documentation should cover questions such as intended context of use, generalizability, performance, robustness, and clinical applicability, at a level of detail sufficient for comprehensive assessment. Specific and clearly formulated regulatory and scientific questions are strongly encouraged to allow reciprocally concise answers.

Applicants should also provide a scientific base with sufficient technical details to allow assessment of AI/ML systems used, data integrity, and model generalizability to the target population and specific context of use.

Source: EMA — Reflection Paper on the use of Artificial Intelligence (AI) in the medicinal product lifecycle (July 2023)