Comparing and Contrasting Individual-Based and Continuum Models of Within-Host Tuberculosis

Abstract Spatial structure and stochastic cellular interactions can influence outcomes of within-host tuberculosis (TB) infection, but their effects may be represented differently in discrete and continuum modelling frameworks. We develop an idealised stochastic individual-based model (IBM), based on a previously published within-host TB model, and formally derive a mechanistically related deterministic continuum counterpart from its probabilistic rules. To investigate how the two modelling approaches compare, we carry out a simulation-based comparative study between appropriately reduced versions of the two models. The two formulations showed varying degrees of agreement in qualitative clearance and non-clearance outcomes, with agreement generally stronger for non-clearance cases. Even where infection outcomes agreed, substantial differences were observed in temporal dynamics and underlying cellular behaviour. Repeated IBM realisations from the same initial configuration demonstrated additional variability arising from intrinsic stochasticity, while differences between matched continuum and IBM simulations may reflect both stochastic effects and differences in model representation. Initial spatial configuration also influenced infection outcomes across otherwise fixed model conditions. Overall, the continuum and IBM formulations provide complementary perspectives on early within-host TB dynamics, with the continuum model capturing population-level behaviour and the IBM retaining discrete interactions and fine-scale spatial heterogeneity.