HTAi Annual Meeting 2025

June 14, 2025

Evidence Prime participated in the HTAi 2025 Annual Meeting in Buenos Aires, Argentina. Held on 14–18 June, the meeting brought together the international health technology assessment community to consider how evidence and assessment methods need to evolve.

AI and living HTA

On 16 June, Artur Nowak, co-founder and CTO of Evidence Prime, contributed to a symposium on AI and living HTA alongside Roche. The discussion examined how artificial intelligence could support a more continuous approach to evidence generation and synthesis, and whether existing HTA systems are ready for that change. Our coverage of the symposium provides further context.

Artur Nowak at HTAi 2025 in Buenos Aires.
Artur Nowak at HTAi 2025 in Buenos Aires.

Validation studies and real-world use cases

In his conference retrospective, Artur noted that AI for evidence synthesis was a central topic throughout the meeting. What he particularly valued was the emphasis on validation studies and real-world use cases: attendees wanted evidence of how these approaches perform in practice.

That emphasis matters for living HTA. New studies continue to appear after an assessment is completed, so keeping the evidence current requires a repeatable process for identifying, reviewing, and incorporating updates. The complete workflow needs to be assessed, including how reviewers check outputs and decide when new evidence changes an assessment.

Practical requirements for continuous assessment

Artur's presentation addressed reproducibility, protection of confidential information, and audit trails that distinguish AI assistance from human actions. It also considered access to full-text publications and the copyright and licensing questions that arise when those publications are processed by AI.

The symposium connected these practical requirements with the wider question of how new methods can fit into established assessment processes. Read the organizer’s meeting review for the broader conference themes.

Blank white image with no visible content or details