How to simulate a drug concentration-time curve with confidence intervals?

How to simulate a drug concentration-time curve with confidence intervals as shown in the figure? It comes from a model shared on File Exchange (ADC Krogh Cylinder Model). I wish to learn what is the process for calculating the confidence intervals? Is it done by generating 100 virtual patients' model parameters that follow a normal distribution based on the mean and CV% of parameters, then simulating the drug concentrations for each patient, and finally calculating the confidence intervals based on the mean and standard error of the concentrations at each time point? Or is there a specific SimBiology function that can directly simulate this type of drug concentration-time curve with confidence intervals by inputting the mean and coefficients of variation of parameters?

Answer 1

[Arthur Goldsipe]

Where did this plot come from? Without more information, we won't be able to say exactly how it was created.

However, it looks like this might be a prediction confidence interval plot created by SimBiology. You can find more information about that functionality on the sbiopredictionci reference page. As for the details of how the confidence intervals are calculated, you will see that the function has two ways of calculating the cofidence intervals, described under the "Type" name-value argument. The faster but less accurate approach is the Gaussian approximation, the slower but more accurate approach is the bootstrap method.

[Shawn]

Hi, Arthur. This plot comes from Figure 3 of the paper Weddell, J., Chiney, M.S., Bhatnagar, S., Gibbs, J.P. and Shebley, M. (2020), Mechanistic Modeling of Intra-Tumor Spatial Distribution of Antibody-Drug Conjugates: Insights into Dosing Strategies in Oncology. Clin Transl Sci. (https://doi.org/10.1111/cts.12892).

[Arthur Goldsipe]

In that case, I suggest reaching out to the corresponding author if the paper doesn't provide sufficient details. From my quick skim of the paper, I couldn't find the details you requested.

[Luis B. Walter]

yes, was done using Simbiology. According to the Methods section of the paper: "A mechanistic ADC PK/clinical response model was developed using the Simbiology application in MATLAB version 2017a (Mathworks, Natick, MA)."

Luis B

[Shawn]

@Arthur Goldsipe Is it the only input way of parameter variation in this command estimatedInfo which is displayed by "Bounds"? If the parameter variation is dispalyed by CV%, should it be transfomed to “bounds” at first?

[Arthur Goldsipe]

Now that you've shared the link to the paper, I think confidence interval calculations (as done by sbioparameterci) will not reproduce this figure. But let me answer your question anyway...

The bounds in estimatedInfo only indirectly control the parameter variation during fitting (and also during the fitting step of bootstrap confidence interval calculations). They are limits that are applied during fitting. Whether or not those bounds actually have any impact on the results depend on a lot of factors.