< Back to Article

Bayesian calibration of a stochastic, multiscale agent-based model for predicting in vitro tumor growth

Fig 7

Parameter sweep.

Effect of proliferation rate (; panel A), death rate (; panel B), glucose uptake (λ; panel C), death rate increase due to lack of glucose (γ_D; panel D), and glucose threshold (σ_H; panel E) for live and dead cell phenotypes. The live cells are seeded with 0.5 confluence, and the dead cells with 0.3 confluence. The base value of the parameters are the ones shown in Table 1 (from the deterministic values column). We changed one parameter at a time by 15% (reducing or increasing up to 30%). According to Panel A, the live and dead cell confluences are directly proportional to the proliferation rate. In Panels B, C, D, and E (i.e., for the death rate, glucose uptake, death rate increase due to lack of glucose, and glucose threshold, respectively), the live cell confluence is inversely proportional to the parameters being tested, while the dead cell confluence is directly proportional.

doi: https://doi.org/10.1371/journal.pcbi.1008845.g007