Why the color of bar plot of elementary effects is so light?

Hello, everyone

I used simbiology for global sensitivity analysis and selected elementray effects method. After obtaining the bar chart, the colors of mean and standard deviation are very light, as shown in the figure below. How to adjust the colors?

Thank you very much.

Answer 1

[Florian Augustin]

Hi,

There is a way to change the color, but I would like explain those bar plots a little. They color-code a histogram of the sensitivities over time (more intense colors mean that the sensitivities occur more often over time, fainter colors mean that this sensitivity occurs less often over time). The sensitivities for outputs 1, 4, and 9 seem to be concentrated around a common value accross some time span, hence you see a darker spot in the bar plots. For global sensitivity bar plots in SimBiology, you can change the color using name-value arguments as documented here (or use the Property Editor of the plot in the SimBiology Model Analyzer if you are using this app). However, this will not change the opacity. For time-varying sensitivities, I'd recommend using the plot function of the sensitivity results. This will show you the sensitivities (mean and standard deviations of elementary effects) over time, which gives a more complete picture of the sensitivities of the outputs over time.

I hope this helps,

-Florian

[voZnW]

Thank you very much for your detailed explanations.

I also want to ask, for time-varying sensitivities,how to set the stop time? Is it based on the elimination time of drugs? For example, if the concentration of the drug cannot be detected after 10 hours, the stop time can be set to 10 hours. If the concentration of the drug can be detected after 1000 hours in the body, the stop time should be set to 1000 hours?

Will the setting of stop time affect the results of sensitivity analysis? If so, in what way will it affect?

Will the setting of stop time affect the running time of sensitivity analysis software?

Thank you very much!

Best wishes for you!

[Florian Augustin]

Hi,

in general, sensitivity results in SimBiology are reported at all output times. This is why looking at time plots of sensitivities is more informative than the bar plots for time varying model outputs; bar plots in turn are better to visualize sensitivities of scalar observables (see examples in second paragraph below). Sensitivities of model responses can change significantly over time. In your case, it seems like the sensitivities of the drug are negligible at times < 1000 hours, but increase from that point on. You should be able to see this transition from "no sensitivity" to "sensitivity" in the time course plots. The output times only need to be large enough to cover the time span of interest.

I am not sure if this is of interest in our analysis, but you can also look at scalar sensitivities, e.g. of a concentration in steady state, or of the Cmax, ... . In this case, the stop time has to be large enough to ensure the system is in steady state (or Cmax has been attained). For example, assuming that the system is in steady state at 1500 hours, then you can set the output times to "1500 hours". If you are interested in the Cmax, then you can define a SimBiology observable and use this as the sensitivity output in the Sobol analysis.

I hope this helps.

Best,

Florian

Answer 2

[Sulaymon Eshkabilov]

x = 1:11;
y = [75 91 105 123.5 131 150 179 203 226 249 281.5];
bar(x, y, 'r')