Introduction

AI tools are increasingly common in pharmacometrics, with the potential to support routine tasks and make complex workflows more efficient [1]. However, systematic
validation of their performance in real-world workflows remains limited [2]. This is especially important for technical AI tools, where outputs may look correct but fail
in practice. A reusable validation procedure was developed using reference examples and automated checks. Two AI tools were systematically validated and benchmarked to assess their reliability and reproducibility in MonolixSuite.

By Stephanie Kollmann, Frano Mihaljevic, Monika Twarogowska

ACoP 2026, October 11-14, 2026, National Harbor, Maryland