CPRED

[DimitrisVoul]

Hello,

I am fitting some data with focei and using xpose to plot the results. Is it possible to use PRED instead of CPRED which is the default choice in the xpose?

and similary WRES instead of CWRES.

[mattfidler]

CPRED is better than PRED; CWRES is better than WRES, so xpose uses over other metrics even with NONMEM.

However, if you want to use these metrics you can:

library(nlmixr)
library(xpose)
#> Loading required package: ggplot2
#> 
#> Attaching package: 'xpose'
#> The following object is masked from 'package:nlmixr':
#> 
#>     vpc
#> The following object is masked from 'package:stats':
#> 
#>     filter
library(xpose.nlmixr)
#> 
#> Attaching package: 'xpose.nlmixr'
#> The following object is masked from 'package:nlmixr':
#> 
#>     vpc

one.compartment.saem <- function() {
    ini({
        tka <- .5   # Log Ka
        tcl <- -3.2 # Log Cl
        tv <- -1    # Log V
        eta.ka ~ 1
        eta.cl ~ 2
        eta.v ~ 1
        add.err <- 0.1
    })
    model({
        ka <- exp(tka + eta.ka)
        cl <- exp(tcl + eta.cl)
        v <- exp(tv + eta.v)
        d/dt(depot) = -ka * depot
        d/dt(center) = ka * depot - cl / v * center
        cp = center / v
        cp ~ add(add.err)
    })
}

fit <- nlmixr(one.compartment.saem, theo_sd, est="saem", control=list(print=500))
#> Compiling RxODE equations...
#> done.
#> 1:    0.1503  -3.3314  -0.6447   0.9500   1.9000   0.9500   9.6167
#> 500:    0.4546  -3.2153  -0.7848   0.4413   0.0713   0.0180   0.4774
#> Calculating residuals/tables
#> done.

xp <- xpose_data_nlmixr(fit,pred="PRED")
absval_res_vs_pred(xp, res="WRES")

^{Created on 2019-02-19 by the reprex package (v0.2.1)}

[mattfidler]

The key is to specify pred when converting nlmixr to xpose and res when calling the specific plots.

[DimitrisVoul]

That's great, thank you very much.

The issue is that I get some negative values with CPRED which makes the plot really difficult to interpret.

Any ideas as to why that might be the case?

[mattfidler]

Hi @DimitrisVoul

I forgot to answer you; I do not know why CPREDs may be negative in this case. I will say that CPRED is closer to what FOCEi optimizes, so it is usually a better metric.

You can also try npdes to see how the model is performining.

f <- nlmixr(model,data,"focei",table=tableControl(npde=TRUE))