Enhancement: NCA Parameter Input

[js3110]

Description

There are many possible parameters available with the PKNCA package. We want an input table in settings for the user to select which parameters they would like to have calculated.
Preferably, parameters that cannot be calculated with the uploaded dataset are removed/hidden from this input table.

Definition of Done

• User can select parameters they want to calculate
• If parameter is not possible to calculate (for example bioavailability or urine) it either does not show up in the input table, or is exlcuded from results.
• App should not crash if unavailable parameter is selected

[js3110]

@billdenney do you have any suggestions for how we can implement this?
At the moment we have a hard coded list of parameters (found in R/format_data.R). The intervals are calculated for each USUBJID and then this list of parameters are set to 'TRUE' .
However, ideally we want all possible parameters to be calculated that the user can then decide to ignore if they want. Is there a way of setting manual start and end times, but default/all parameter calculation?

[billdenney]

That's not the way that PKNCA is intended to work. Rather than calculate everything and then have the user select what they want, the way that PKNCA works is that the user selects what they want and then the required parameters are calculated. Some parameters cannot be calculated for every interval (for instance, IV parameters should not be calculated for extravascular data).

The preferred implementation would be to select the desired set of parameters for calculation based on the types of data you're currently analyzing (proposed in billdenney/pknca#157). This is how it's done in WinNonlin and OpenNCA. The reason that PKNCA doesn't directly follow the design that others have used is that it prevents analysis of multiple types of data at the same time. For instance, if you have both IV bolus, extravascular, and urine data in the same dataset, PKNCA can analyze it all at the same time. With WinNonlin, you have to run 3 separate times and then pull the data back together.

The difference by implementing the PKNCA issue above is that it would be applied per-interval. So, the power to calculate many different types of data simultaneously would be maintained.

Additionally, some parameters (like time_above) require user input of how they want the parameter calculated, so it's not possible to just calculate "something" for that parameter.

[js3110]

Thanks Bill! Thats kind of what I meant; I know PKNCA can't calculate all the parameters for every interval, so I was wondering if there is a way for pknca to calculate all of the parameters that it is able to, and leave the ones that it can't. That way the user never has to select anything and it is all done automatically. If that makes sense.

I have some follow up questions:

1. You said that currently pknca can calculate parameters for different data at the same time, so that means you can define intervals/parameters for different types of data at the same time too right? For example IV parameters. And then presumably for the EV subjects and data these parameters won't be calculated? Do they have to be explicitly set in the intervals or is pknca able to automatically detect this?
2. Do you have any documentation regarding each parameter and what data they need to be calculated? I know it can be found for each individual function in the docs but I'd rather ask incase you already have it summarised somewhere 😅

:)