Discrepancy between model fits and simulations.
I am fitting a generic TMDD model to date. Model fits look reasonable. Once I create a variant and simulate data for various doses to create observed vs predicted concentrations vs time profile, then simulated concentrations do not match fitted profiles. Simulated concentrations are either significantly higher or lower than the model fits. Not sure what is wrong. Fits look fine but simulated profiles using fitted parameters are all over the place.
2 Comments
Time Descending
Does the data that you use for fitting include dosing? If so, perhaps you are not applying the dosing in the same way when doing simulations.
If this explain the discrepancy, is there any way you can share your code with us? It's quite hard to debug these sorts of issues when we can't reproduce the problem.
Are you applying the doses using dose objects in SimBiology or by changing the initial conditions? If the latter, a problem could be that you might have initial conditions that are non-zero. Doses in SimBiology are additive so if - during the fitting process - the initial condition C(t=0) = 10 and you add a bolus dose of 100 to C at t = 0, the value of C at t=0 will be 110. However, if you subsequently simulate the model by changing the initial conditions, the non-zero C(t=0) = 10 will not be added to the dose and the value of C at t=0 would be 100. I hope that makes sense.
