COSBI at BIOMATH 2026
Quantitative Systems Pharmacology (QSP) models play an increasingly important role in understanding disease mechanisms, predicting treatment responses, and supporting model-informed drug development. A key element of many QSP applications is the generation of virtual populations, which enable researchers to represent biological variability and uncertainty in mechanistic simulations. However, generating realistic virtual populations often requires a large number of model evaluations, making the process computationally expensive.
At BIOMATH 2026, held in Szeged, Hungary, COSBI PhD student Marco Bozza presented the poster “Surrogate-Assisted Virtual Population Generation in QSP Models.”
The work investigates the use of Gaussian Process surrogate models to accelerate the generation of plausible patients and virtual populations. Rather than relying exclusively on repeated simulations of computationally intensive mechanistic models, the proposed approach uses surrogate representations of model outputs to efficiently guide parameter selection and identify candidate virtual patients.
The methodology was evaluated on QSP models of increasing complexity and compared with established virtual population generation approaches, including Simulated Annealing, Nested Simulated Annealing, and Markov Chain Monte Carlo methods. The results demonstrate that surrogate-assisted strategies can substantially reduce computational costs while maintaining comparable performance in reproducing target biological variability and preserving model goodness-of-fit.
As mechanistic models continue to increase in scale and complexity, computational efficiency becomes a critical factor for their practical application. This work contributes to the development of scalable methodologies that can facilitate the broader adoption of QSP approaches in biomedical research and pharmaceutical development.
The study was conducted by Marco Bozza, Stefano Giampiccolo, and Luca Marchetti, reflecting COSBI’s ongoing commitment to advancing computational methods for systems pharmacology and model-informed drug development.
Explore COSBI publications, posters and scientific outputs.
Digital copies of selected posters are available upon request at info@cosbi.eu.
