Question : PKNCA.options specific to certain profiles

[js3110]

Hi! I'm looking into the PKNCA.options function as part of working on issue #148 of aNCA, and what we would ideally want is the option to change the PKNCA.options parameters, most specifically the allow.tmax.in.half.life , but only for specific profiles. For example if the half life calculation is manually inputted, we would want the tmax.in.half.life to be TRUE, but in all other cases we want it to be FALSE.

Currently I cannot see a way to customise this for each profile, but maybe this is something that can be implemented? similar to the way that the intervals can be customised per profile as well?

[billdenney]

If the half-life is manually specified with include_half.life, then the option excluding tmax is ignored because the user has manually chosen the points. So, I think that what you're requesting for the effect exists now (while the specific request about different options per interval does not). Does this cover your use case and if not can you please clarify the use case more (since I think that this does cover the use case)?

I hesitate to make the options differ by interval because they are intended to be methodology relating the the entire analysis rather than per-interval selections. (When I'm writing reports, I use the options as part of automatic report generation in the methods section. If options were per-interval, then it would be very difficult to do that.)

See the example below where lambda.z.time.first is 2 without include_half.life and is 1 with it.

library(PKNCA)
#> 
#> Attaching package: 'PKNCA'
#> The following object is masked from 'package:stats':
#> 
#>     filter
d_conc <- data.frame(conc = c(0, 2^(0:-4)), time = 0:5, hl_include = c(FALSE, rep(TRUE, 5)))

o_conc <- PKNCAconc(d_conc, conc~time)
o_data <- PKNCAdata(o_conc, intervals = data.frame(start = 0, end = Inf, half.life = TRUE))
o_nca <- pk.nca(o_data)
#> No dose information provided, calculations requiring dose will return NA.
as.data.frame(o_nca)
#> # A tibble: 10 × 5
#>    start   end PPTESTCD            PPORRES exclude
#>    <dbl> <dbl> <chr>                 <dbl> <chr>  
#>  1     0   Inf tmax                 1      <NA>   
#>  2     0   Inf tlast                5      <NA>   
#>  3     0   Inf lambda.z             0.693  <NA>   
#>  4     0   Inf r.squared            1      <NA>   
#>  5     0   Inf adj.r.squared        1      <NA>   
#>  6     0   Inf lambda.z.time.first  2      <NA>   
#>  7     0   Inf lambda.z.n.points    4      <NA>   
#>  8     0   Inf clast.pred           0.0625 <NA>   
#>  9     0   Inf half.life            1      <NA>   
#> 10     0   Inf span.ratio           3      <NA>

o_conc_incl <- PKNCAconc(d_conc, conc~time, include_half.life = "hl_include")
o_data_incl <- PKNCAdata(o_conc_incl, intervals = data.frame(start = 0, end = Inf, half.life = TRUE))
o_nca_incl <- pk.nca(o_data_incl)
#> No dose information provided, calculations requiring dose will return NA.
as.data.frame(o_nca_incl)
#> # A tibble: 10 × 5
#>    start   end PPTESTCD            PPORRES exclude
#>    <dbl> <dbl> <chr>                 <dbl> <chr>  
#>  1     0   Inf tmax                 1      <NA>   
#>  2     0   Inf tlast                5      <NA>   
#>  3     0   Inf lambda.z             0.693  <NA>   
#>  4     0   Inf r.squared            1      <NA>   
#>  5     0   Inf adj.r.squared        1      <NA>   
#>  6     0   Inf lambda.z.time.first  1      <NA>   
#>  7     0   Inf lambda.z.n.points    5      <NA>   
#>  8     0   Inf clast.pred           0.0625 <NA>   
#>  9     0   Inf half.life            1      <NA>   
#> 10     0   Inf span.ratio           4      <NA>

^{Created on 2025-02-24 with reprex v2.1.1}

[js3110]

Ok, I see, upon further investigation I think we have a bug in our package which is why this isn't working for me! You are right, this should cover the use case. I just have two follow up questions:

1. If the user inputs a manual slope selection, this allows tmax to be included, but does this also keep the previous option of FALSE for the rest of the profiles? So only the profile that was selected is changed.
2. Does this also apply to min.hl.points? For example if only 2 points are added as inclusions, but min.hl.points is set to 3, will it still calculate using that, or will it give an exclude message?

I'll have to dive deeper into it on my end to figure out why its not working in the app .

Thanks!

[js3110]

Also to clarify, this has been requested as something our scientists would like to do to explore the data, but not to use for reporting. They just want to be able to see what different slopes might look like.

[billdenney]

If you use the include_half.life option, it overrides everything. You can choose two points, the points can have an increasing slope, the user can do whatever they want (for better or for worse).

You should set profiles without manual point selection to NA rather than FALSE for automatic point selection. FALSE means to exclude the point from selection while NA means to use automatic methods. This is a recent modification, so you will need to install the version from GitHub rather than CRAN to use varying manual half-life point selection.

For revising point selection and dynamically updating the NCA, that is something that was recently requested in #385. The goal there was to make the NCA calculation responsiveness faster in ruminate which would likely have the same goals as aNCA for responsiveness. You may want to watch that issue for when it is implemented.

[js3110]

yes an update function would be even more ideal, so I will defo keep an eye on that! Thanks