1 reply to this topic

[Teodora]

Hi,

I am working on an indirect response model. As a first step I am trying to model only the response data using the following code:
test(){

EFF=0
sequence{E = 1.35} # initializes the PD compartment to a literature value

deriv(E = Kin*(1-EFF) - Kout * E)

Kin=Kin0*exp(-Kdel*t) # assume Kin decreased exponentially with time

stparm(Kdel = exp(tvKdel + nKdel))
stparm(Kin0 = exp(tvKin0 + nKin0))
stparm(Kout = exp(tvKout + nKout))

fixef(tvKdel = c(, -4.81, ))
fixef(tvKin0 = c(, 1.99, ))
fixef(tvKout = c(, -3.69, ))

error(EEps = 0.2)
observe(EObs = E*(1+ EEps))

ranef(block(nKdel,nKin0, nKout)= c(1,0.15,1,0.24,0.05,1))
}


As you see, in the absence of predose values for the biomarkers, I am initializing the compartment with a value from the literature. Although the model fit is nice, I would like to add some variability to this value, so that different subjects have different initial PD compartment values.

I tried the following code:

stparm(base=exp(baseTV + nbase))
fixef(baseTV(freeze) = 0.3) # (log of 1.35)
ranef(diag(nbase) = 0.2))

sequence{E = base}

However, after running the model, all subjects turn out with identical IPREDs at time zero (with a value within the variability set above).

Do you have any suggestions?

Thanks,
Dora

[serge guzy]

Dear Dora

You have one syntax error

 

ranef(diag(nbase) = 0.2))

 

 

should be

 

 

ranef(diag(nbase) =c( 0.2))

 

 

Hope it helps

Best

Serge