Hello, I’m trying to fit free and bound drug concentrations using the protein binding isotherm in a Phoenix Model. Unlike normal explicit equations, I can’t get Phoenix to recognize “t” as the time. I have the appropriate data column mapped to “time” in the initial mapping (free drug). I started from a non-population PK model, and tried to amend code. Here is what I have so far, 479 is the estimated Pt from graphing: test(){ C = NK479t/(1+Kt) error(CEps = 1) observe(CObs = C + CEps) stparm(N = (tvN)) stparm(K = (tvK)) fixef(tvN = c(, 1, )) fixef(tvK = c(, 1, )) } I’ve tried using the old t = x trick, specifying time in the Model Inputs, and nothing seems to be working. I sorted the data from smallest to largest values of free drug (x or time). Thanks, Julie
Hi Julie, Add this line to your code: covariate(t) Then go map “t” to your time column in the main panel. Backgrounder: time, denoted as “t”, is implicitly added to models using the closed form equations (cfMicro, cfMacro, etc…) or “deriv” statements. Otherwise “t” is undefined, and needs to be declared as a covariate, parameter, or a LHS (left hand side) value in an equation. For additional functionality with respect to plots, modify your observe statement so that Phoenix knows t is an independent variable as well: observe(CObs(t) = C + CEps) Happy Modeling, -Jason
Thank you! It seems to be working fine for the modeling, but the IPRED vs. IVAR plots aren’t. I just made them in Excel- should they work in Phoenix with the Cobs(t) statement? I am using the multiplicative error model, if that makes a difference.
The error model itself won’t make a difference. It looks like there’s a bug in that setting “t” as the independent variable doesn’t work. Maybe there is an internal conflict with the name. When I use “x” instead of “t” I can get the plots.