Proposal Description

Recent advances in emulation have made it possible to estimate models/parameters that balance process understanding with the behaviour observed in the data from real systems. These hybrid approaches to modelling offer powerful insights across disciplines such as health, environmental science, engineering, and the life sciences, opening the door to models that both explain and predict complex phenomena. As inference methods have matured, attention has shifted towards understanding and quantifying the uncertainty that surrounds such hybrid estimates, which is essential for assessing robustness and ensuring that model-based conclusions remain credible.

Uncertainty quantification presents real challenges, particularly for high dimensional, computationally intensive, and sensitive to underlying assumptions. Nevertheless, new statistical and computational approaches are beginning to address these difficulties. This session highlights recent progress in uncertainty quantification for structured models/emulators, with four talks showcasing different techniques and application areas to provide a clear and up-to-date view of how uncertainty can be characterised, propagated, and managed in practice.