Hi, I am trying to obtain bootstrap parameter estimates from a PKPD model. I have runned the bootstrap and it works fine. However, I havent been able to stratify my bootstrap. In my bootstrap estimates, I would like to have “GROUP” as covariate but when the software doesnt allow me to stratify my bootstrap. “GROUP” is a covariate in the text file. The covariate is in text format. I tried to use a numerical covariate as well but neither does that work. See caption from text file below. In the attached figure you can see how it looks when I try to select stratification. I currently use Phoenix WinNonlin 6.2 and NLME 1.1. What is missing here to do a stratified bootstrap? Kind regards, Rasmus test(){ covariate(tvKa, tvV1,tvCL,tvCLD,tvVd,Salt,Bio,GROUP) # PK deriv(Aa = - tvKa * Aa) deriv(A1 = tvKaAa - tvCLC- tvCLDC +tvCLDC2) deriv(A2 = tvCLDC - tvCLDC2) #Disease model deriv(E = Rate* (1-(1*C/(IC50+C)))) 
Dear Rasmus Rather than stratify by GROUP, I suggest you to use GROUP as a covariate in the model. The code would be test(){ cfMicro(A1, Cl / V) dosepoint(A1) C = A1 / V error(CEps = 0.0891822) observe(CObs = C * (1 + CEps)) stparm(V = tvV * exp(dVdGROUP2*(GROUP==2)) * exp(nV)) stparm(Cl = tvCl * exp(dCldGROUP2*(GROUP==2)) * exp(nCl)) covariate(GROUP()) fixef(tvV = c(, 47.5205, )) fixef(tvCl = c(, 5.38279, )) fixef(dVdGROUP2 = c(, 0.00048101, )) fixef(dCldGROUP2 = c(, -0.0546043, )) ranef(diag(nV, nCl) = c(0.086794151, 0.060539687)) } You will get CL and V at each bootstrap for GROUP 1 and for group 2 by x tvV by exp((dVdGROUP2) for V and similarly for Cl (Cl instead of V). I think that there is an issue with the bootstrap stratification although you have a code error to specify GROUP as covariate. You should have additional () to specify GROUP is a categorical covariate. Although this would let you do stratification, I do not see in the outputs the impact of GROUP. Therefore, I used directly GROUP as a covariate in the model as in the project attached. At least you can retrieve theta for each GROUP at each bootstrap. best Serge covariate(tvKa, tvV1,tvCL,tvCLD,tvVd,Salt,Bio,GROUP()) PS: You should get NLME1.2, this is the official version. [file name=bootstrap_official.phxproj size=739521]Certara | Drug Development Solutions (722 KB)