Related Experiment Videos
Interchangeability and predictive performance of empirical tolerance models
M Gårdmark1, L Brynne, M Hammarlund-Udenaes
1Department of Pharmacy, Uppsala University, Sweden.
Clinical Pharmacokinetics
|March 27, 1999
Summary
Empirical tolerance models often lack mechanistic insight. This study found that while multiple models can fit data, their predictive power depends heavily on the original study design, not just model type.
Area of Science:
- Pharmacology
- Mathematical Modeling
- Systems Biology
Background:
- Tolerance models are frequently empirical due to complex physiological mechanisms.
- Selecting appropriate models and data analysis strategies for tolerance is challenging.
Purpose of the Study:
- To evaluate the interchangeability of various empirical tolerance models.
- To identify determinants for selecting suitable tolerance model designs and data analysis strategies.
Main Methods:
- Seven diverse empirical tolerance models were selected.
- Data were simulated from each model and analyzed using all other models.
- Model performance, predictive capacity, and influence of study design were assessed.
Main Results:
- Each dataset was adequately described by at least two models, but no single model fit all datasets.
- Model predictive capability correlated with original study design, not model type.
- Extensive tolerance data could only be described by a limited number of models.
Conclusions:
- Reliable mechanistic insights from tolerance data are unlikely without additional knowledge.
- Study design is crucial for the predictive validity of tolerance models.
- Factors influencing the design and evaluation of tolerance studies were identified.