An efficient and flexible framework for inferring global sensitivity of agent-based model parameters.

Area of Science:

• Computational Biology
• Systems Biology
• Ecological Modeling

Background:

• Agent-based models (ABMs) simulate complex systems, but their predictions require uncertainty quantification via global sensitivity analysis.
• Existing global sensitivity methods are computationally expensive, limiting their application to complex ABMs.

Purpose of the Study:

• Introduce SMoRe GloS (Surrogate Modeling for Recapitulating Global Sensitivity), a computationally efficient method for ABM global sensitivity analysis.
• Enable accurate uncertainty quantification and parameter space exploration for complex ABMs.

Main Methods:

• Leverage explicitly formulated surrogate models to approximate ABM behavior.
• Apply SMoRe GloS to in vitro cell proliferation and 3D vascular tumor growth ABMs.
• Compare SMoRe GloS performance with Morris and eFAST methods.

Main Results:

• SMoRe GloS achieved substantial speedups, completing analyses in minutes compared to days for eFAST.
• The method accurately recovered global sensitivity indices for both simple and complex biological ABMs.
• SMoRe GloS estimated sensitivities for parameters not explicitly in the surrogate model.

Conclusions:

• SMoRe GloS offers a computationally efficient and accurate solution for global sensitivity analysis in ABMs.
• This method enhances uncertainty quantification and model reliability for complex systems.
• SMoRe GloS facilitates deeper exploration of model behavior and increases confidence in predictions.