NA value is available when I conduct simulation in loop

[huyo1444]

Hi.

I wanted to conduct simulation with some sets of PK parameter value.
so I created parameter set in csv file and conducted simulation with following described code,
there are no problem, and run code, but there are some NA values in output.

for (i in 1:2) {
dat <- expand.ev(amt = parameters[i,5], ka = parameters[i,7], Fa = parameters[i, 6],
ii = 24, addl = 9)
out <- mod %>%
data_set(dat) %>%
Req(CP, Xgut) %>%
mrgsim(end = 240, delta = 1, out = "df")
out$ID<-i
output<-bind_rows(output,out)
}

[kylebaron]

Hi @huyo1444 -

It's hard to tell where the NA might be coming from (the input, the model or the output processing). I've reproduced your original problem showing inputs and outputs. Could you do the same? I've also included a re-work of the problem that would be more efficient solution and reduce chances of processing problems.
Thanks,
Kyle

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(mrgsolve)
#> 
#> Attaching package: 'mrgsolve'
#> The following object is masked from 'package:stats':
#> 
#>     filter

Reproduce original problem

nothing <- c(0,0)
parameters <- data.frame(
  a = nothing, 
  b = nothing, 
  c = nothing, 
  d = nothing, 
  e = c(100, 200), # amt
  f = c(1,0.5), # Fa
  g = c(1, 1.1) # ka
)

parameters
#>   a b c d   e   f   g
#> 1 0 0 0 0 100 1.0 1.0
#> 2 0 0 0 0 200 0.5 1.1

code <- '
[ param ] CL = 1, V = 20, ka = 1, Fa = 1
[ main ] 
F_cent = Fa;
double KA = ka;
[ pkmodel ] cmt = "Xgut cent", depot = TRUE
[ error ] capture CP = cent / V;
'

mod <- mcode("na", code)
#> Building na ... done.

output <- data.frame()

for (i in 1:2) {
  dat <- expand.ev(amt = parameters[i,5], ka = parameters[i,7], Fa = parameters[i, 6],
                   ii = 24, addl = 9)
  out <- mod %>%
    data_set(dat) %>%
    Req(CP, Xgut) %>%
    mrgsim(end = 240, delta = 1, out = "df")
  out$ID<-i
  output<-bind_rows(output,out)
}

head(output)
#>   ID time       Xgut       CP
#> 1  1    0   0.000000 0.000000
#> 2  1    0 100.000000 0.000000
#> 3  1    1  36.787944 3.070263
#> 4  1    2  13.533528 4.050011
#> 5  1    3   4.978707 4.268005
#> 6  1    4   1.831564 4.212711
summary(output)
#>        ID           time            Xgut                 CP        
#>  Min.   :1.0   Min.   :  0.0   Min.   :  0.00000   Min.   : 0.000  
#>  1st Qu.:1.0   1st Qu.: 59.0   1st Qu.:  0.00000   1st Qu.: 3.725  
#>  Median :1.5   Median :119.5   Median :  0.00061   Median : 5.265  
#>  Mean   :1.5   Mean   :119.5   Mean   :  9.46260   Mean   : 5.898  
#>  3rd Qu.:2.0   3rd Qu.:180.0   3rd Qu.:  0.27207   3rd Qu.: 7.454  
#>  Max.   :2.0   Max.   :240.0   Max.   :200.00000   Max.   :12.517

You could make this more efficient with

data <- mutate(parameters, amt = e, ka = g, Fa = f, time = 0, evid = 1, ID = c(1,2), 
               ii = 24, addl = 9, cmt = 1)

output <- mod %>% 
  data_set(data) %>%
  Req(CP,Xgut) %>% 
  mrgsim(end = 240, output = "df")

head(output)
#>   ID time       Xgut       CP
#> 1  1    0   0.000000 0.000000
#> 2  1    0 100.000000 0.000000
#> 3  1    1  36.787944 3.070263
#> 4  1    2  13.533528 4.050011
#> 5  1    3   4.978707 4.268005
#> 6  1    4   1.831564 4.212711

^{Created on 2021-12-12 by the reprex package (v2.0.1)}

[huyo1444]

Hi Kyle

Thanks to replay my issue and very helpful suggestion,
I tried to your code, but it could not improve my work....

My first question is not good, and I corrected it.
In my output, NA value was available, when I use "spread" and recreate output table.

It seemed that, mrgsolve was not return all required time point (in spite of I set "delta = 1")
and, returned time point was different in each "ID". So if I used "spread" key = ID.

please advice again.
#####################################################################
for (i in 1:66) {
dat <- expand.ev(amt = parameters[i,5], ka = parameters[i,7], Fa = parameters[i, 6],
ii = 24, addl = 9)
out <- mod %>%
data_set(dat) %>%
Req(Cp, Xgut) %>%
mrgsim(end = 240, delt = 1, out = "df")
out$ID<-i
output<-bind_rows(output,out)
}

output <- output %>%
dplyr::filter(Xgut>0)

output <- output[,-4]
output <- spread(output1, key = ID, value = Cp)

[kylebaron]

please advice again.

Provide the reproducible example that I asked for in the first reply.

It seemed that, mrgsolve was not return all required time point (in spite of I set "delta = 1")
and, returned time point was different in each "ID".

You'll have to provide a reproducible example for this too; write some code that I can run and share it here that shows this behavior.

Thanks,
Kyle

[huyo1444]

Dear Kyle

Thanks to replay my asking.
I'm sorry, my issue was resolved.

The issue was occurred because of simple code miss.
(I conducted filter the data with "Xgut>0, and lost some data. set filter function Xgut!=0, it's resolved)

Thanks for your counsel.

huyo1444