Session (CTI)

Purpose Health economic models are designed to forecast health and cost outcomes associated with a specific treatment in a particular disease area. Technical specifications for such a model must be sufficiently detailed to serve as a blueprint for model programming. From the blueprint, the model can be implemented in Excel or coded in R. We will demonstrate how to leverage agentic AI to go from model concept to full implementation in both R and Excel. Description Workshop attendees will get working knowledge on the use of agentic AI to generate both the blueprint and the implemented economic model. The workshop will show how AI, with human in the loop, is used for the blueprint development and then for implementation. A detailed example in autoimmune disease will be used throughout. Participants will also receive a link to a site from where they can download the Excel engine, templates and the full example. [Jaime] will moderate the session, introducing the problem and guiding the active participation from the audience. [Brian] will describe the humans’ role in conceptualizing the model and introduce the autoimmune disease example. [Baris] will explain how a multi-agentic AI system with structured prompting can be used to develop, review and edit the specifications based on a modelling paper to generate the blueprint and then how a complementary agentic system can generate the corresponding R code, run analysis and develop the technical report. Then [Apoorva] will show how the same blueprint can be used with a general Excel template to create the spreadsheet version of the same model. [Jaime] will then lead the interaction with the participants, aided by probing questions and polling. This workshop will be valuable to researchers interested in leveraging agentic AI to create health economic models. The methodology of structured prompting with automated templates also can help with many other research tasks in the HEOR space.