RE: Simulate with ETAs from .phi and residual variability

Dear Philp, You may also want to have a look at the open-source R package campsis ( https://calvagone.github.io/). Below is some R-code that illustrates how you can easily simulate individual profiles taking the ETAs from a NONMEM output table. Best regards, Andreas. --------------------------------------------------------------------------------------------------------------------------- library(tidyverse) library(campsis) # Load a standard model from the library # you can also set-up a custom model using the free online simulator e-campsis.com) model <- model_library$advan2_trans2 # set the parameter values model <- model %>% replace(Theta("V", value=20)) model <- model %>% replace(Theta("CL", value=5)) model <- model %>% replace(Omega("V", value=0.1,type = 'var')) model <- model %>% replace(Omega("CL", value=0.1,type = 'var')) model <- model %>% replace(Sigma("PROP", value=0.02,type = 'var')) # Have a look at the model model # load NONMEM output table containing the ETAS (and covariates if you have) ETA.tab <- read_table("patab001",skip = 1) %>% group_by(ID) %>% slice(1) %>% # take only the first record of each subject rename(ETA_CL=ETA1,ETA_V=ETA2) # you may need to rename variables to match with the model code # set-up the simulation schedule; the number of subject must equal that of ETA.tab dataset <- Dataset(subjects = n_distinct(ETA.tab$ID)) %>% add(Bolus(time=0, amount=1000, ii=12, addl=2)) %>% add(Observations(times=seq(0, 36, by=0.5))) # add the ETAs by using the bootstrap functionality dataset <- dataset %>% add(Bootstrap(data= ETA.tab, id="ID", replacement=FALSE, random=FALSE, export_id=TRUE)) # we need to discard the random IIV as the ETAs are taken from the dataset model@parameters@list <- model@parameters@list %>% purrr::discard(.p=~is(.x, "omega")) # run the simulation results <- model %>% simulate(dataset=dataset,outvars = c('BS_ID')) # BS_ID is the original ID in the ETA.tab # plot the results spaghettiPlot(results,'OBS_CP')