NCA: Allow selection of points on individual profiles to censor, use in half-life, etc

[john-harrold]

After running NCA the individual profiles are generated. Add the ability to select points and flag them for

• exclusion/inclusion from analysis
• exclusion/inclusion for half-life estimation
• Any others?

@billdenney This seems to describe how to manually select points:

https://billdenney.github.io/pknca/articles/v06-half-life-calculation.html#manual-point-selection

Is there a way to flag rows to ignore or should the dataset just be filtered first? I'm asking because it might be good to have those points available in the output just marked as excluded (for post processing).

[billdenney]

Well, you can't flag a row for general inclusion-- just exclusion.

General exclusion removes a value from all calculations. Half-life exclusion removes a value from half life calculations, and automatic point selection is used. Half life inclusion includes the specified values and does no automatic half-life point selection.

You are correct that it's best to keep the values in the dataset. The way to use those functions is to add a column indicating if the value should be included or excluded and give that column name to PKNCA.

[john-harrold]

I have a question about the include_half.life option. If I select points for only one subject for inclusion, will the other subjects behave as though no points for inclusion were selected and use the curve stripping option?

[john-harrold]

Oh I missed it in the warnings. So it seems if any points are selected for inclusion it expects that all points will be specified. Is that correct?

[john-harrold]

Hey @billdenney one other question. Currently I'm generating tables of individual observations. I'm thinking that I should tag those with the flags the user has applied: N.NN^{HLINC} or N.NN^{CENS} where N.NN is the actual observation like 2.45 or something. And then tables with HLINC or CENS would have a note indicating: HLINC: Manually selected to include in hal-life determination, CENS: Manually censored from NCA. Does that sound useful?

Currently BLQ are just that, no number. I'd keep those the same.

[billdenney]

The intent of the manual point specification is that you can specify it for some of the points (TRUE or FALSE in the column) or allow automatic selection (NA). And, this should apply for an entire interval. But, you should be able to mix automatic and manual between different intervals.

[billdenney]

And, yes, creating a listing of concentration time data with exclusions or other choices would be helpful.

[john-harrold]

The intent of the manual point specification is that you can specify it for some of the points (TRUE or FALSE in the column) or allow automatic selection (NA). And, this should apply for an entire interval. But, you should be able to mix automatic and manual between different intervals.

So if I have an interval where some points are NA and some are TRUE how is that handled for that interval?

[billdenney]

With a new issue in PKNCA. 😄

I think that the way it should be handled within PKNCA is that the include_half.life column should be all TRUE or FALSE within one interval. It should never be a mix of TRUE/FALSE with NA within an interval. It should be fine to be a all NA for an interval (but it currently isn't).

[john-harrold]

Can I request that it can be TRUE, FALSE, or NA in an interval and that if TRUE or FALSE values are specified the the NAs get converted to FALSE? it will make flagging things easier for me.

[billdenney]

Yes, can do.

[billdenney]

Fixed in the development version of PKNCA.

[john-harrold]

Fixed in the development version of PKNCA.

Cool. I'll try it out tomorrow.

[john-harrold]

And, yes, creating a listing of concentration time data with exclusions or other choices would be helpful.

I know it's not latex :), but how does this look?

[billdenney]

That looks good. The time column should sort numerically rather than lexigraphically. How would it work if you have more subjects than would fit on one page?

[john-harrold]

Good point. I'll fix the sorting. I'm using the span table function below and it will chop up a large table into smaller tables such that they fit with constraints on width, height, number of rows, and number of columns.

That table is actually split into 8 smaller tables.

https://onbrand.ubiquity.tools/reference/span_table.html

[john-harrold]

Fixed in the development version of PKNCA.

This works. Thank you!

[billdenney]

Great!