A Nonparametric Approach to Practical Identifiability of Nonlinear Mixed Effects Models

Area of Science:

• Mathematical Biology
• Computational Biology
• Biostatistics

Background:

• Mathematical modeling is key for clinical trial data interpretation.
• Individual-based fitting is common, but hierarchical approaches are increasingly used in pharmacometrics.
• Existing parameter identifiability techniques are challenging to apply in hierarchical settings.

Purpose of the Study:

• To propose a novel nonparametric method for studying practical identifiability.
• To address the limitations of current identifiability techniques in hierarchical parameter estimation.
• To demonstrate the utility of the proposed method in nonlinear mixed-effects modeling.

Main Methods:

• Developed a nonparametric approach to assess practical identifiability.
• Focused on the nonlinear mixed-effects (NLME) framework.
• Applied the method to two established examples from pharmacometrics and viral dynamics.

Main Results:

• The proposed nonparametric method is effective for studying identifiability in hierarchical models.
• Demonstrated the approach's applicability and potential utility.
• Provided insights into parameter identifiability within complex modeling frameworks.

Conclusions:

• The nonparametric approach offers a valuable tool for analyzing parameter identifiability in hierarchical models.
• Facilitates more robust interpretation of clinical trial data in pharmacometrics and viral dynamics.
• Advances the understanding and application of hierarchical parameter estimation.