352

---

Click the image below to view the full size version of this cover.

Issue 352: Agent-based modeling is an important tool for computational modeling of cellular processes. However, classical analytical tools are difficult to bring to bear on these models due to their large number of parameters, stochasticity, and computational complexity. Surrogate models, computationally efficient models that capture the dominant features of a complex system, offer a means to illuminate these otherwise unobservable features of agent-based models (ABMs). Our new method leverages explicitly formulated surrogate models to unlock global sensitivity analyses of ABMs. By connecting the ABM to the surrogate model in our SMoRe GloS framework, we can infer sensitivity metrics of the features targeted by the surrogate model to variations in the underlying ABM parameter space. This approach reveals how subtle changes in model parameters influence emergent behaviors, offering new avenues for understanding cancer systems.