ISoP ISoP on Twitter | ACoP ACoP on Twitter | PAGE | WCoP | PAGANZ | PAGJa

Modeling Initial Conditions in Monolix without using ODEs


#1

Good afternoon everyone,
I have PopPK data where the participants have plasma levels at time=0.

The three compartment model file is:

[LONGITUDINAL]
input = {Cl, V1, Q2, V2, Q3, V3}

EQUATION:
V = V1
k = Cl/V1
k12 = Q2/V1
k21 = Q2/V2
k13 = Q3/V1
k31 = Q3/V3

Cc = pkmodel(V, k, k12, k21, k13, k31)

OUTPUT:
output = Cc

However, the various entries I tried has not worked properly. What do you recommend?

Thanks,
Andy


#2

Dear Andy,

as this model is linear, if there is a non 0 value of Cc at time 0 you just have to add a Cc_init parameter and put another equation Cc_out = Cc+Cc_init

  • Cc_init can be a constant value
  • Cc_init can be an input and defined as a regressor
    Notice that Cc = pkmodel(V, k, k12, k21, k13, k31) +Cc_init will lead to an error as the pkmodel macro should be “alone” on its line.

Hope this helps

Jonathan


#3

Hi Jonathan,
I did what you stated, but the model would not compile.

[LONGITUDINAL]
input = {Cl, V1, Q2, V2, Q3, V3, Cc_init}
regressor = {IC} ; values at time zero

EQUATION:
V = V1
k = Cl/V1
k12 = Q2/V1
k21 = Q2/V2
k13 = Q3/V1
k31 = Q3/V3

Cc = pkmodel(V, k, k12, k21, k13, k31, Cc_init)
Cc_out = Cc + Cc_init

OUTPUT:
output = Cc_out

Any thoughts?

Andy


#4

I was more thinking at

LONGITUDINAL]
input = {Cl, V1, Q2, V2, Q3, V3, Cc_init}
Cc_init = {use=regressor} ; values at time zero

EQUATION:
V = V1
k = Cl/V1
k12 = Q2/V1
k21 = Q2/V2
k13 = Q3/V1
k31 = Q3/V3

Cc = pkmodel(V, k, k12, k21, k13, k31)
Cc_out = Cc + Cc_init

OUTPUT:
output = Cc_out

It should work with the 2016R1 (otherwise the syntax of the regressor part is a little bit different)

Jonathan


#5

Hi Jonathan,
Thanks for the modification…based on what you sent me for version 2016R1, here is the error message I get when I try to compile the model:

<img src="/upload

Description: Incomplete match, remaining unmatched input is:
Then it shows the model you typed….

Andy


#6

Dear Andy, I guess [LONGITUDINAL] is missing

I tired with a simpler case, with the theophylline example.
-> I change the data set and copy the ID column to create a new column IDcopy to create a regressor.
-> For the library model, I created oral1_1cpt_kaVCl_reg.txt as the following

[LONGITUDINAL]
input = {ka, V, Cl, Cc_init}
Cc_init = {use=regressor}

EQUATION:
Cc = pkmodel(ka, V, Cl)
Cc_out = Cc+Cc_init

OUTPUT:
output = Cc_out

When I clicked on “check initial fixed effects”, I can see that the initial condition is well take into account.

Best


#7

Hi Jonathan,
It’s a winner…that is fantastic…the model works well by estimating the initial conditions using the regressor function.

This was outstanding…thanks for your help Jonathan…

Andy


#8

Great

Enjoy MonolixSuite2016R1

Best