Presentation (CTI)

OBJECTIVES: Manufacturers increasingly use artificial intelligence (AI) tools, including large language models (LLMs), to support evidence synthesis, systematic literature reviews, and health economic modelling within HTA dossier submissions. This study aimed to map existing guidance addressing manufacturer AI use in dossier preparation and identify accountability gaps between industry obligations and HTA agency expectations.
METHODS: A rapid targeted literature review was conducted across guidance documents and policy statements from ISPOR, HTAi, EMA, NICE, HAS, CDA, and FDA published between 2022 and 2026. Databases and sources searched included PubMed, ISPOR.org, agency websites, and grey literature repositories. Documents were screened for content addressing AI disclosure obligations, validation requirements, and accountability allocation in the context of HTA dossier submissions.
RESULTS: The review identified no HTA agency with an explicit published policy governing AI use by manufacturers within dossier submissions. The ISPOR ELEVATE-GenAI reporting guidelines (2025) addressed methodological transparency for LLM use in HEOR but did not assign accountability between manufacturers and reviewers. The HTAi Global Policy Forum (2026) acknowledged mixed views on GenAI adoption but produced no submission-specific recommendations. Key unaddressed areas consistently identified across sources included: validation requirements for AI-generated evidence, mandatory disclosure obligations, and liability allocation when AI-assisted outputs contain errors.
CONCLUSIONS: A critical accountability vacuum exists at the intersection of manufacturer AI use and HTA review. As AI adoption in dossier preparation accelerates, the absence of clear governance standards creates risks for submission integrity and reviewer confidence. These findings call for urgent collaborative development of submission-specific AI governance guidance involving manufacturers, HTA agencies, and AI tool developers.