PMX-CovEval: A Framework Including a Simulated Pharmacokinetic Database for Covariate Model Building Methods...

Area of Science:

• Pharmacometrics
• Pharmacokinetics
• Computational Biology

Background:

• Covariate model building (CMB) is crucial in population pharmacokinetics (popPK) for understanding drug variability.
• Existing methods for CMB lack a standardized framework for evaluation and benchmarking.
• This necessitates the development of a unified resource to compare different CMB strategies.

Purpose of the Study:

• Introduce PMX-CovEval, a comprehensive framework for evaluating CMB methods in popPK.
• Provide a standardized and reproducible resource for benchmarking various CMB techniques.
• Facilitate systematic comparison of CMB strategies in real-world pharmacokinetic modeling.

Main Methods:

• Developed PMX-CovEval, a framework comprising 127 diverse scenarios.
• Included pharmacokinetic (PK) datasets, model files for NONMEM and Monolix, and empirical Bayes estimates (EBEs).
• Designed scenarios to reflect real-world complexity while ensuring practical usability.

Main Results:

• PMX-CovEval offers a unified resource with PK datasets, model files, and EBEs.
• The framework supports evaluation of standard CMB techniques available on PsN and Monolix.
• Initial benchmarking results are provided to demonstrate the framework's utility.

Conclusions:

• PMX-CovEval establishes a standardized, reproducible framework for CMB method evaluation in popPK.
• This resource will aid researchers in systematically comparing and selecting optimal CMB strategies.
• The framework promotes advancements in covariate model building for improved pharmacokinetic analyses.