I have some questions about the analysis of sparse samples using NCA and NLME.
In NCA there is an option “sparse” to choose but I didn’t find a similar one in NLME. How does NLME know whether these are sparse samples or not?
In NCA for sparse samples nominal time is used for analysis. But if times randomly selected from a set of times for each region, not fixed times for each region, are used, should we use nominal time or actual time?
Is there any difference between the data format of sparse samples and that of “normal” samples?
In NCA there is an option “sparse” to choose but I didn’t find a similar one in NLME. How does NLME know whether these are sparse samples or not?
NLME is designed to analyze sparse data (popPK). Sometimes to build a relevant pop model for rich dataset is even harder
In NCA for sparse samples nominal time is used for analysis. But if times randomly selected from a set of times for each region, not fixed times for each region, are used, should we use nominal time or actual time?
I would suggest to use nominal times. Please see the User’s guide about PK metric calculation in sparse design. WNL calculates the mean for each point. So if some of your subjects was sampled at 24.1h, it would be another point for AUC calculation etc.
Is there any difference between the data format of sparse samples and that of “normal” samples?
there is no difference. If you have PHX1.4 installed please see the NCA example of sparse data (I don’t know why a few useful datasets were removed from PHX7)
In NCA there is an option “sparse” to choose but I didn’t find a similar one in NLME. How does NLME know whether these are sparse samples or not?
NLME is designed to analyze sparse data (popPK). Sometimes to build a relevant pop model for rich dataset is even harder
Dose this mean that if we have rich dataset and the built-in PK library contains the right model, we prefer built-in model to NLME?
In NCA for sparse samples nominal time is used for analysis. But if times randomly selected from a set of times for each region, not fixed times for each region, are used, should we use nominal time or actual time?
I would suggest to use nominal times. Please see the User’s guide about PK metric calculation in sparse design. WNL calculates the mean for each point. So if some of your subjects was sampled at 24.1h, it would be another point for AUC calculation etc.
For example, we collected 2 samples from each of the regions: 0.5-1h, 1-2h, 2-3h and 3-4h postdose. How to determine the nominal times?
Dose this mean that if we have rich dataset and the built-in PK library contains the right model, we prefer built-in model to NLME?
How do you define which model is right?
In your opinion what is the difference between built-in PK library and NLME?
For example, we collected 2 samples from each of the regions: 0.5-1h, 1-2h, 2-3h and 3-4h postdose. How to determine the nominal times?
It is not what I meant.
Nominal time is if you have some subjects who should be sampled say at the time 2h (nominal time). But the real sampling time was different (+/- 5 min for example) - that’s actual time.
