tsk_estimation(algorithm = "fo")

[SCarter79]

Using R 4.0.05 and assemblerr e452c176. When building a 1 compartment oral model and then using tsk_estimation(algorithm) = "fo" , the model renders ok, but the ETAs are not 0 FIX:

prazmod <- pk_model() +
  dataset(here('DATA.csv')) +
  pk_absorption_fo() +
  pk_distribution_1cmp() + 
  pk_elimination_linear() + 
  obs_combined(conc~C['central'])

render(prazmod,
       tasks = tsk_estimation(algorithm = "fo") 
       )

gives the following output:

$PROBLEM
$INPUT ID AGE RACE HT HCTZ AMT TIME DV RATE EVID SS II SID OCC
$DATA DATA.csv IGNORE=@
$SUBROUTINES ADVAN2  TRANS2 
$PK
MAT = THETA(1) * EXP(ETA(1))
VC = THETA(2) * EXP(ETA(2))
CL = THETA(3) * EXP(ETA(3))
KA = 1/MAT
V = VC
$ERROR
CONC = A(2)/VC
Y = CONC + EPS(1) + CONC * EPS(2)
$ESTIMATION METHOD=0 MAXEVAL=999999
$THETA (0, 0.5, Inf) ; POP_MAT
$THETA (0, 100, Inf) ; POP_VC
$THETA (0, 50, Inf) ; POP_CL
$OMEGA 0.1 ; IIV_MAT
$OMEGA 0.1 ; IIV_VC
$OMEGA 0.1 ; IIV_CL
$SIGMA 1; RUV_ADD
$SIGMA 0.1; RUV_PROP

[sebastianueckert]

FO does not imply fixing the variance to 0. However, having another estimation method "pooled" is a very good idea for the future.