Issue with PK PArameters calculated by SA

[msevestre]

see https://github.com/Open-Systems-Pharmacology/OSPSuite.ReportingEngine/issues/218

[msevestre]

@Yuri05 This is in fact by design. For SA, we are explicitly excluding the norm parameters Probably because we are using Dose and BW which are also parameters to calculate the norm parameters and therefore the SA based on this output would be wrong?

@pchelle and @adullahhamadeh What are your thoughts on this? Does it make sense to calculate a SA value using input parameters of the model other than the one being varied?

Especially when varying volume or bodyweight. The value we would use would be wrong as we would use the value of the BW in the simulations and not after variation for a specific SA run

see https://github.com/Open-Systems-Pharmacology/OSPSuite.Core/blob/develop/src/OSPSuite.Core/Domain/Services/SensitivityAnalyses/SensitivityAnalysisPKAnalysesTask.cs#L15

[Yuri05]

ok, I see the point. When calculating pk values in SA (or in a Pop simulation): currently Dose and BW are taken from the basis simulation and not from the simulation with varied parameters; thus all _norm parameters would be wrong as soon as dose and/or bw are not constant during variation. That's why _norm parameters are not calculated in SA (and also return NaN in Pop simulations).

To do it properly also for _norm parameters, in https://github.com/Open-Systems-Pharmacology/OSPSuite.Core/blob/develop/src/OSPSuite.Core/Domain/Services/PKAnalysesTask.cs basis simulation should be updated with individual parameters before calculating pk values (and resetted at the end).

For V9 it's too late to change this.

My suggestion for now would be:

• OSPSuite.Core remains AS IS
• Reporting Engine:
  • Parameter names passed to all PK-dependent tasks (calculate PK, plot PK, calculate SA, plot SA) are the same (=Status Quo, nothing to change)
  • "calculate PK/plot PK"-Task throws an error if PK-parameter name is invalid (=Status Quo, nothing to change)
  • "calculate SA/plot SA" tasks show just a warning if PK-parameter name is invalid and skips this parameter for plots (this would be the case only if PK-parameter name is valid as such but not returned by core SA calculation; thus currently for all _norm parameters)

Thoughts?

[msevestre]

There is no basis simulation. We calculate the pk solely based on the output values.

In PKSim, we do calculate pk values for pop by overriding the value of bodyweight every time. This is not something we do for SA whatsoever.

I don't think it's straight forward with the current infrastructure

[Yuri05]

There is no basis simulation. We calculate the pk solely based on the output values.

Basis simulation is the one used for parameter variation https://github.com/Open-Systems-Pharmacology/OSPSuite.Core/blob/f5bf2c15ab385fe5bcd6f900e3d1f9c4795861a2/src/OSPSuite.Core/Domain/Services/SensitivityAnalyses/SensitivityAnalysisRunResultCalculator.cs#L35

It's also passed to PKAnalysisCalculator but as far as I can see is not used later on to get updated BW/Dose values for each "Individual" (= parameter variation in SA context)

Anyway, I would not change it in Core for V9.

[msevestre]

Sure. But it's already using the results. The variations were applied outside . Definitely no way to change that in v9