Agent-based models (ABMs) are powerful tools for exploring how individual interactions lead to complex, emergent behaviors in biology, ecology, and society. Yet, assessing the uncertainty of these models can be prohibitively time-consuming. SMoReGloS (Surrogate Modeling for Recapitulating Global Sensitivity) introduces a breakthrough approach that dramatically reduces the computational cost of global sensitivity analysis. By using surrogate models to efficiently explore parameter space, SMoReGloS delivers accurate and comprehensive sensitivity estimates in minutes rather than days. Tested on both simple and complex biological models—including 3D tumor growth simulations—SMoReGloS matched the accuracy of traditional methods like eFAST while offering remarkable speedups. This innovation makes in-depth uncertainty quantification feasible for even the most computationally demanding ABMs, empowering researchers to better understand and trust their model predictions.
