A have a few questions for the experts. Any help is greatly appreciated. I’m using the classic version of Phoenix NLME 1. Is it a good practice to use your PK estimates as link paramters for you PK-PD model? 1. What are the best criteria for evaluating a linked model? 2. I’m getting results from my linked model, but should I be concerned if the core ouput also contains the warning “Input units do not allow for conversion” for most subjects? Is this likley due to the fact I’m using beats per min (bpm) as units for my heart rate effect data? Should this instead be represented as a % of the baseline values? 4. Is there a way to obtain the mean, median, range, etc., for all secondary and final parameters between all six subjects? Thank you Marc R.
Hi Marc! [quote]1. Is it a good practice to use your PK estimates as link parameters for you PK-PD model?[/quote]Why not? Unless you deal with an endogenous compound or measure some PD endpoint which will feedback on the PK (very unlikely - at least after a single dose) PD is driven by PK. Variability of PD is generally higher than PK. Quite often you have to apply different weighting schemes (PK covering a wide range whereas blood pressure does not). You can fix parameters from the PK fit in the PKPD model in a first step and try to allow PK parameters to vary in a second step. [quote]1. What are the best criteria for evaluating a linked model?[/quote]The same ones as for PK. Visual inspection (!), residual plots, min. AIC/SBC (matter of taste). Not important IMHO: correlation, SSQ. [quote]2. I’m getting results from my linked model, but should I be concerned if the core ouput also contains the warning “Input units do not allow for conversion” for most subjects? Is this likley due to the fact I’m using beats per min (bpm) as units for my heart rate effect data?[/quote]The warning should be given for all subjects.. [quote]Should this instead be represented as a % of the baseline values?[/quote]Wouldn’t help. But nothing to worry about. [quote]4. Is there a way to obtain the mean, median, range, etc., for all secondary and final parameters between all six subjects?[/quote]Since you are dealing with classical (individual) modeling you get one set of parameters for every subject. “Final parameters Pivoted” and “Secondary Parameters Pivoted” > NCA and Toolbox > Descriptive Stats