Quiz 4. Model Development

What is the purpose of the base model?

• To test covariates

• To estimate variability on bioavailability

• To define the structural PK model and variability

True or False: You can still change the number of compartments after you’ve started covariate inclusion

• True

• False

Answer explanation

False – once the base model is locked in, structural changes should stop.

What is the minimum drop in OFV required to consider a covariate significant at p < 0.05 during forward inclusion?

When are you most likely to apply allometric scaling?

• During final model

• During intermediate model

• During base model

• During complete model

True or False: AIC is used to compare non-nested models

• True

• False

What’s the rule of thumb when deciding when a covariate is adequately represented in your dataset?

• At least 50% of the population must have it

• You need at least 10 individuals per category

• At least ~20% of the population should fall within the covariate group of interest

• Covariates can always be tested, regardless of distribution

Which of the following would be appropriate for screening continuous covariates?

• Boxplots

• Correlation analysis

• Chi-square test

Answer explanation

Correlation analysis is correct for continuous covariates.
Boxplots are more relevant for categorical covariates.

What OFV increase corresponds to p < 0.01 during backward elimination?

• 10.83

• 3.84

• 6.63

Which of the following best describes the complete model?

• Model with only continuous covariates

• Model with all significant covariates

• Model without covariates

• Model after validation 

What should the final model be used for?

• Predicting drug concentrations

• Performing dosing simulations

• Dose optimisation

• Supporting clinical interpretation and decision-making