Hi,
In the diagnostic plot (in this case popDV vs IPRED), is it possible to identify which dot belongs to what data point? And is it also possible to give dots a colour based on e.g. gender? That would help to identify certain groups that are not well caught with the model.
Esther
hi Esther, yes this is possible, if you click on the point inthe plot (can be ‘delicate’) it will pop up the source datasheet used to make the plot with the point hightlighted
Looks good in your example. I tried it, the window is flickering as if it is doing something, but the data source won’t show up. Can it be the version I use? I use Phoenix version 8.5.
no - it should be the same for most versions; I mouse over the point of interest then double click. I’m at the EU UGM today but happy to have a short call with you tomorrow if you’re’ still stuck?
The double clicking did the trick, thanks.
Is it also possible to give certain dots in these graphs a colour, so colour by group (gender, dose etc)?
i think maybe if you email an example project to support we can pick this up and look at your specific usecase - most likely i would propose informing the model that doseLevel, gender etc. are potential covariates then they are available in the output for subsequent transformation or plotting
you cannot color by say outlier value, what you can do is use your results worksheet take it into a custom data wizard to say identify the points you want to highlight and then do a custom plot where you color by these values. if you are using a pop pk model we usually care about CWRES not WRES also beware of the zero zero time point this is not a data point we never observe a zero it might be a BLQ. your DV is 2443 while IPRED is 1000 it can be a wrong dose or a wrong time but 2 is not bad usually we use CWRES ±4/5 as benchmark for outliers
