OBJECTIVES: The use of artificial intelligence (AI) promises a future where health technology assessment (HTA) processes can be faster and more efficient. Currently, there are no frameworks on the use and acceptability of AI-based solutions by HTA agencies. The aim of this study is to explore the extent to which HTA agencies can prepare for integrating AI in their decision-making process.

METHODS: Semi-structured, qualitative interviews were conducted with eight stakeholders from diverse professional functions, including experts on HTA process and decision-making, real-world evidence, health economics, literature reviews, and advanced analytics, to gather their insights on AI integration into HTA. A thematic analysis of data was conducted, with support of AI-based software, to identify key expectations, preferences, and challenges regarding the integration of AI in HTA.

RESULTS: Experts expect technical guidance on AI-based solutions to vary across HTA agencies. As guideline development may not be a priority for HTA agencies, disparities in timings for AI integration across agencies is expected. The main technical challenges for incorporating AI-based solutions include potential bias in analyses outputs, variable reproducibility, and lack of transparency, together with abundance of AI-based solutions, lack of in-house technical expertise, and geopolitics. Given the lack of existing guidelines, the challenges identified, and the need to maximize efficiencies, experts expect HTA agencies to initially integrate AI-based solutions into literature reviews and economic modelling, which involve high computational power and human labor.

CONCLUSIONS: AI-based solutions have the potential of maximizing human resource rationalization and institutional efficiency, providing support for both manufacturers and HTA agencies. HTA agencies are expected to focus primarily on the transparency, reliability, and reproducibility of AI-generated evidence. While AI-based solutions are expected to become a tool, the ultimate decision-maker will remain the HTA stakeholders. Consequently, in-house AI expertise will still need to be accompanied with expertise on HTA drivers and processes for decision-making.