Presentation (CTI)

OBJECTIVES: Dossiers are comprehensive, evidence-based documents that support payer and health technology assessment (HTA) decision-making but are often resource-intensive to develop. Generative artificial intelligence (GenAI) can accelerate the process by automating evidence synthesis and content generation, improving efficiency and consistency. This systematic literature review (SLR) summarizes published evidence on the application of AI/machine learning (ML) tools in dossier development.
METHODS: EMBASE^®, MEDLINE^® and ISPOR databases were searched from inception to present to identify studies evaluating the integration of AI/ML in the process of dossier development. The review followed the standard methodology for conducting SLR as per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.
RESULTS: Overall, 12 studies met the inclusion criteria. Most studies assessed the use of diverse AI tools to support global value dossier (GVD) creation process, while two studies specifically utilized AI to develop country‑specific dossiers for Germany and Canada. The most commonly used technologies were GenAI and Large Language Models (LLMs), with specific tools including LiveRef™, LiveSTART™, DO-BO®, and Coauthoring Accelerator. These tools were primarily applied in three key areas of GVD: content generation, data extraction, and document drafting. AI-assisted process resulted in substantial time savings, ranging from 53.0-99.8% (n=9), compared to the manual approach. Accuracy was consistently high across most studies, ranging from 76-95% (n=4). Only one study utilizing the DO-BO® tool reported cost savings of €78,000. Importantly, majority of the studies emphasized the necessity of a human-in-the-loop approach, wherein subject matter experts retained responsibility for validation, refinement, and final approval of AI-generated outputs to ensure outline completeness, flow and relevance.
CONCLUSIONS: This SLR highlights the evolving role of AI to transform dossier creation process, offering substantial time and resource savings while maintaining high accuracy and quality. However, human oversight remains essential, particularly for ensuring strategic alignment, complex data visualization, and final quality control.