Fig. 3: Comparison between the experimentally measured tumor volume and inference of logistic growth with a linear treatment term (Linear Treatment Model, Eq. (3)) for each treatment protocol.
From: Modeling tumor dynamics and predicting response to therapies in a murine pancreatic cancer model
Panels A–E correspond to the following treatment protocols: A) NGC, B) NGC+ Losartan, C) NGC + Calcipotriol, D) Calcipotriol, and E) NGC + Calcipotriol + Anti-PD-L1 mAb. The dashed black lines are the line of unity. All parameter estimations to treatment scenarios exhibit high levels of correlation between the data and model, with all CCCs and PCCs greater than 0.98. The average accuracy, calculated as the number of correctly identified responders and non-responders divided by the total number of mice for each treatment, was 98.89 ± 1.11% across all treatment scenarios. Further, the MAPE for allscenarios is less than 10%. This suggests that the Bayesian estimation of model parameters (r, α, and N0) effectively reproduces the experimental data.
