Parameter estimation with sbioparmestim, using experimental data replicates or averages / standard deviations

Toolboxes: Simbiology, Global Optimization Toolbox

Function: sbioparamestim

Issue: Want sbioparamestim to estimate parameters using data replicates or averages & standard deviations

Example Data A:

tspan = [0 10 30 60 100]';

xtarget = [0 0.35 0.4 0.48 0.39]';

sbioparamestim works well with dataset A. xtarget has one set of values to match. However, if my data were to contain replicates:

Example Data B:

xtarget_rep_1 = [0 0.3 0.4 0.48 0.39]';

xtarget_rep_2 = [0 0.3 0.4 0.36 0.39]';

extarget_rep_3 = [0 0.3 0.4 0.2 0.39]';

I do not know how to integrate such data into a parameter estimation function. Optimization Toolbox and Simbiology documentation did not cover such examples to my knowledge.

Answer 1

[Arthur Goldsipe]

Hi,

sbioparamestim does not currently support fitting data that contains replicate measurements. To use sbioparamestim, you would need to preprocess the experimental data to only have on observation per time point (for example, by averaging the replicates).

If your goal is to fit the data, another option would be to use the sbionlinfit function, which supports data that contains replicates. Note that this function relies on the Statistics Toolbox function nlinfit. This function is designed around fitting pharmacokinetic data sets, so before you can estimate the data you will need to a PKModelMap object. You might find it helpful to review the following example that shows how to fit a model using both sbioparamestim and sbionlinfit:

http://www.mathworks.com/help/simbio/examples/simulating-the-glucose-insulin-response.html