Dosing EVIDs (1 or 4, and maybe others) without an AMT column should give a warning (or error?)

[billdenney]

library(nlmixr2)
#> Loading required package: nlmixr2data
library(nlmixr2lib)
library(babelmixr2)

d_simple <-
  data.frame(ID = 1, EVID = c(1, 0), cmt = c("depot", "central"), DV = c(NA, 1), TIME = 0:1)

model <- readModelDb("PK_1cmt")

babelmixr2:::.bblDatToNonmem(model = model, data = d_simple, table = nlmixr2est::tableControl(),
                             fun="bblDatToPknca", replaceEvid=5L,
                             replaceOK=TRUE, software="pknca", env=NULL)
#>   ID TIME EVID AMT II DV CMT DVID SS nlmixrRowNums
#> 1  1    0    2   0  0 NA   1    0  0             1
#> 2  1    1    0  NA  0  1   2    0  0             2

^{Created on 2023-05-30 with reprex v2.0.2}

[billdenney]

Actually, it is not specific to linCmt(). I don't know why EVID 1 is becoming EVID 2.

[billdenney]

The underlying issue was that there was no AMT column.

So, that changes the issue request to please give a warning when there are dosing EVIDs but no AMT. Is there a scenario where that combination could be desired?

library(nlmixr2)
#> Loading required package: nlmixr2data
library(nlmixr2lib)
library(babelmixr2)

d_simple <-
  data.frame(ID = 1, EVID = c(1, 0), cmt = c("depot", "central"), DV = c(NA, 1), TIME = 0:1, AMT = c(1, 0))

model <- readModelDb("PK_1cmt")

babelmixr2:::.bblDatToNonmem(model = model, data = d_simple, table = nlmixr2est::tableControl(),
                             fun="bblDatToPknca", replaceEvid=5L,
                             replaceOK=TRUE, software="pknca", env=NULL)
#>   ID TIME EVID AMT II DV CMT DVID SS nlmixrRowNums
#> 1  1    0    1   1  0 NA   1    0  0             1
#> 2  1    1    0  NA  0  1   2    0  0             2

^{Created on 2023-05-31 with reprex v2.0.2}

[mattfidler]

AMT could be missing with a $PRED model. EVID really should be missing there too....

[billdenney]

@mattfidler, Yes, I can see scenarios where AMT is missing (like a $PRED-only model), and I can even imagine that EVID could be present in those cases (e.g. you want to use EVID = 2 to get a prediction without an observation or to modify some time-dependent phenomenon). But, I can't see a scenario where EVID indicates a dose and AMT is missing (or NA).

That's making me think that EVID indicating a dose and no AMT column should be an error.

[mattfidler]

Probably. This is now in the wrong repository. Transferring.

[mattfidler]

FYI, amt=na were already handled.