95% Confidence Interval: How To Display Results?

Open-Systems-Pharmacology / PK-Sim

PK-Sim® is a comprehensive software tool for whole-body physiologically based pharmacokinetic modeling

Other

106 stars 50 forks source link

95% Confidence Interval: How To Display Results? #1253

Open NicolasATX opened 5 years ago

NicolasATX commented 5 years ago

Hello everyone,

I am new to PKSim and I have #an issue that is certainly simple to solve. I built a model for large molecule with TMDD interaction. In this particular model I want to estimate two parameters:

Reference concentration of binding partner (as I have in vitro data for Koff and Kd, I didn't want to leave ref. conc. to default 1 µM value)
Kd FcRn in endosomal space: in this case I have contradictory in vitro data and want to estimate it.

The fitting is quite good with observed data but when I run a Parameter Identification in order to estimate parameters' optimal value and distribution, in the Covariance Matrix tab, the 95% CI is not reported for reference concentration (see capture). The value may be calculated but the format may prevent good display. Do you have any idea how I could access to this data?

Nicolas

tobiasK2001 commented 5 years ago

hm, strange. Just a quick check. Did you just try to increase the number decimal places via: Than go to your PI and create a new covariance matrix. How does it looks like?

NicolasATX commented 5 years ago

Hello Tobias,

Thank you for your answer. I did try to change decimal but still CI is displayed as "0" (no more digits while the mean is displayed with additional digits). For your information I also tried to change units and this is not helping as CI is still displayed as "0". Do you think that CI is effectively calculated and reported as 0? In some cases I saw that CI may not be calculated and are reported - in that case - as "NaN".

Nicolas

tobiasK2001 commented 5 years ago

Dear Niclas,

I could reproduce the issue and I think PK-sim might be not behaving correct here. Could you report your finding here: https://github.com/Open-Systems-Pharmacology/PK-Sim/issues

Thnaks! Tobias

msevestre commented 5 years ago

@NicolasATX I have moved the issue to the PKSim repository since it seems to be a bug However, in order to understand the problem., could you please let us know how to reproduce it?

Attach a project that we can use to recreate the issue
Describe what needs to be done in order to recreate the issue
Describe the issue (e.g. expected results vs what is happening in the tool).

If we have this information, we will be able to understand what is going on and maybe even provide you with a workaround

Cheers, Michael

NicolasATX commented 5 years ago

Thank you Tobias and Michael for your feedback,

Here is a light version of the project. I removed different simulations I previously performed (e.g. without TMDD etc...) to make it more comprehensive. However it appears that with this version, the CI of the estimate 'Reference Concentration' is no more calculated --> 'NaN' is returned instead of '0' in the former version in the covariance matrix tab..

Simulation with TMDD for GitHub.zip

Could you possibly re-run the parameter identification (MonteCarlo + Sensitivity analysis) in order to estimate 'Reference concentration' with a CI?

Best, Nicolas