Session (CTI)

PURPOSE: In the past year, the field of Generative AI (GenAI) has seen remarkable advancements, significantly impacting its applications in HEOR. These advancements offer exciting new opportunities but also demand a rigorous and scientific approach to maintain the integrity and reliability of generated evidence. This session aims to provide attendees with a thorough overview of these recent developments. By addressing both the potential benefits and the challenges, we will delve into the implications for the reliability, transparency, and validity of GenAI in HEOR. Attendees will gain the latest knowledge and insights, ensuring they are well-informed about the current state and future potential of GenAI in HEOR. DESCRIPTION: The session, moderated by Siguroli Teitsson, will begin with a concise summary (7’) of developments in Generative AI (GenAI) over the past year, including the availability of more powerful large language models (LLMs), increased prominence of advanced prompt engineering and publication of guidance and regulation regarding the use of GenAI in HEOR. Sven Klijn from BMS will discuss the advances in prompting techniques and the implications for reliability, transparency and validity of GenAI applications (10’). Tim Reason from Estima Scientific will present on novel use cases for GenAI in HEOR, that were hitherto not possible or lacked scientific robustness (10’). Dr. Rachael Fleurence from NIH will reflect on the rapid progress in the field of GenAI from a policy perspective, with a focus on Health Technology Assessment and the Joint Clinical Assessment (10’).Throughout the presentations, the audience will be interactively polled to gather community perspectives regarding recent GenAI developments. A significant part of the session will be reserved for open discussion, encouraging active audience participation. By facilitating a multi-perspective dialogue, the session seeks to address existing concerns regarding GenAI and identify robust methods of working in this rapidly developing field.