Euro Roundup: EMA researchers share regulatory research priorities for AI across the lifecycle
Researchers from the European Medicines Agency (EMA) have published a paper on regulatory research priorities for AI across the medicine lifecycle.
Published as a preprint, the paper features the findings of a survey of 273 people, including regulators, pharmaceutical industry professionals, patients, academics, and healthcare professionals. EMA and its collaborators ran the survey to understand priority research areas for AI, from product development to post-authorization monitoring and evaluation.
“Accuracy and reliability of AI tools” was the highest priority across all groups “by a substantial margin,” the researchers said. Respondents were most concerned with how AI-generated outputs can be trusted when they inform evidence generation and regulatory decisions. “Data governance, confidentiality, and consent” and “ethics, fairness, and bias prevention” were the next highest priorities.
Based on the findings, the researchers drafted 10 priority research areas. The top three areas all relate to accuracy and reliability, with the researchers wanting to understand how AI systems can remain robust in the face of changing data and when it is important for models to be explainable.
The researchers published the paper one day after EMA released its 2025 AI Observatory report. EMA used the report to provide an overview of AI guidance and policy, applications, collaborations, and engagement activities. The report includes a list of AI uses EMA has discussed with applicants, such as the use of generative AI to draft regulatory submissions or to generate answers to queries.
EMA shared the preprint paper and AI Observatory report as part of the agenda for a meeting scheduled for next week. The one-hour session will cover topics such as approaches for generating evidence on methodologies that enable safe and reliable AI integration. EMA included the session in a regulatory science research meeting that will also cover topics such as new approach methodologies.
EMA Paper, Observatory Report, Draft Agenda
European Commission posts first joint clinical assessment report under HTAR
The European Commission has published a joint clinical assessment (JCA) report on tovorafenib, marking a key milestone in the implementation of the Health Technology Assessment Regulation (HTAR).
Under HTAR, European Union nations jointly evaluate the clinical benefits of new medicines and certain medical devices. The collaboration is intended to reduce duplication, save time, and give companies a clearer and more predictable process. EU Member States will use JCA reports to inform their decisions on which products to make available to patients.
Having started 18 JCAs since HTAR took effect in January 2025, Member States have now completed the first report. The document covers tovorafenib, Ipsen Pharma’s treatment for the most common type of pediatric brain tumor. The Commission granted a one-year conditional marketing authorization to the kinase inhibitor on 20 April.
The JCA report is a 154-page document that describes the medical condition and drug product, assesses the evidence on the treatment, and includes appendices about the joint clinical assessment process. The Member State Coordination Group on Health Technology Assessment also provided a 20-page summary of the report.
The report focuses on the evidence in three populations, namely the full claimed indication and two subsets of patients with a BRAF mutation. Within each population, the assessors reviewed different interventions, comparators, and health outcomes, referred to by the acronym PICO.
Ipsen provided evidence for two of the eight PICOs and argued that it was unable to address four PICOs because of a lack of data on the specified comparators in the target population. The company tried to create external control arms for the four PICOs without relevant data but was unable to find a suitable source.
The assessors found the information insufficient to support assessment of one of the PICOs for which Ipsen submitted results, limiting the report to one of the eight scenarios. For the one PICO with evidence to inform the analysis, the assessors found “major, general uncertainties” relating to evidence synthesis methods, the small sample size, and the differences between the study populations.
MHRA’s AI sandbox project reveals need for new and updated UK guidance
The UK Medicines and Healthcare products Regulatory Agency (MHRA) has named areas where guidance needs creating or updating to address issues identified in its AI Airlock program.
Having finished the second phase of the program, MHRA noted a need for guidance on predetermined change control plans (PCCPs). The guidance should provide “principles-based clarity on the limits of changes that could be included in a PCCP,” MHRA said. Managing iterative updates to AI systems safely and proportionately across the device lifecycle using PCCPs was a focus in phase two of the program.
Guidance on performance metrics for AI-powered in vitro diagnostics (IVDs) is needed too, MHRA said. The planned guidance will include examples and practical approaches and address “the circumstances in which additional metrics add value, and how analytical performance concepts can be adapted for AI systems,” MHRA said.
The agency also proposed updating its software qualification and intended purpose guidance. MHRA wants the “foundational pieces of guidance to clearly address the changing device landscape, including AI-specific considerations and adaptive systems.”
MHRA published the recommendations alongside new simulation workshop reports on IVDs, PCCPs, and the scope of intended purpose. The agency also shared details of a secure sandbox for testing AI-enabled devices in the London healthcare service.
EMA finalizes external control paper, staying on course for consultation in Q4
EMA has finalized its concept paper on using external controls for evidence generation in regulatory decisions.
The agency ran a consultation on a draft version of the paper last year to gather views on its plans to create a reflection paper. EMA argued a reflection paper is needed to address the rise in proposals to use external controls to support regulatory decisions. The trend created a need for general principles for assessing when external controls are appropriate, EMA said.
EMA’s final concept paper features tweaks to the draft released for consultation. The agency expanded the problem statement, adding that terms such as synthetic data and digital twins need clarifying, and noting issues with the replication and reproducibility of results.
Maintaining the timeline established in the draft, EMA’s Committee for Medicinal Products for Human Use aims to discuss the draft reflection paper in the fourth quarter and hold a three-month consultation. EMA plans to finalize and adopt the reflection paper by the second quarter of next year.
EFPIA urges lawmakers to strengthen Biotech Act’s extended protection proposal
EFPIA is pushing EU legislators to strengthen the 12-month supplementary protection certificate (SPC) extension proposed in the Biotech Act.
The trade group welcomed the European Commission’s proposed extension for biotechnology-derived products and advanced therapy medicinal products. However, as the European Parliament starts work on the Biotech Act, EFPIA wants legislators to expand the scope of the extension to all innovative medicines and to establish “clear and predictable award criterias.”
EFPIA backed its request with research showing that the scope and eligibility of the SPC extension will determine its impact. The EFPIA-commissioned research found the extension could attract €45 billion ($52 billion) in inward investment into the EU over 15 years under the broadest scenarios, compared to €4 billion under the narrowest scope. EFPIA predicts the scope will also affect clinical trial enrollment.
