Federico Reali at PAGE 2026
Virtual populations (VPops) are a fundamental component of Quantitative Systems Pharmacology (QSP), enabling researchers to represent biological variability and uncertainty when simulating disease progression and treatment response. As QSP models continue to increase in complexity, however, generating realistic virtual populations can become a significant computational challenge.
At the Population Approach Group Europe (PAGE) Meeting 2026 in Dubrovnik, Croatia, COSBI researcher Federico Reali presented the poster “No One-Size-Fits-All: A Multi-Model Evaluation of Optimization-Based and Surrogate-Assisted Virtual Population Generation in QSP.”
The study investigates different approaches for generating virtual populations, comparing optimization-based methods and surrogate-assisted strategies across three QSP case studies of increasing complexity. The analysis evaluates each method according to multiple criteria, including goodness-of-fit, efficiency, diversity and computational cost.
The results demonstrate that there is no universally optimal strategy for virtual population generation. Instead, performance depends on the specific model and the criteria being prioritized. While surrogate-assisted approaches are not consistently superior across all scenarios, they can provide substantial reductions in computational time as model complexity increases, while maintaining competitive performance.
These findings highlight the importance of selecting virtual population generation methods according to the characteristics of the model, available computational resources and research objectives. By providing a systematic comparison across multiple applications, the work contributes to the development of more efficient and scalable workflows for model-informed drug development.
This research reflects COSBI’s commitment to advancing computational methodologies that support the practical application of QSP models in biomedical research and pharmaceutical development.
Explore COSBI publications, posters and scientific outputs here
Digital copies of selected posters are available upon request at info@cosbi.eu.
