Cluster Gauss-Newton method for a quick approximation of profile likelihood: With application to...

Area of Science:

• Pharmacokinetics and Physiological Modeling
• Computational Biology
• Statistical Modeling

Background:

• Physiologically-based pharmacokinetic (PBPK) models are crucial for drug development but often suffer from parameter uncertainty due to limited observable data.
• Traditional methods for parameter identifiability and confidence interval estimation, like profile likelihood, require extensive computational resources due to repetitive optimizations.
• The Cluster Gauss-Newton method (CGNM) offers efficient parameter space exploration but has not been directly leveraged for profile likelihood approximation.

Purpose of the Study:

• To develop an efficient method for approximating profile likelihoods in PBPK models.
• To reduce the computational burden associated with parameter identifiability analysis.
• To enable rapid estimation of parameter confidence intervals and identify parameter combinations.

Main Methods:

• Proposed a novel approach to approximate profile likelihood by reusing intermediate computational results from the Cluster Gauss-Newton method (CGNM).
• Implemented a technique to derive upper bounds of the profile likelihood without additional model evaluations.
• Extended the method for rapid generation of both one-dimensional and two-dimensional profile likelihoods.

Main Results:

• Successfully approximated profile likelihoods for all unknown parameters in PBPK models with significantly reduced computation time.
• Demonstrated the ability to generate two-dimensional profile likelihoods for parameter combinations within seconds.
• Validated the method's effectiveness across three distinct PBPK models, showing its practical applicability.

Conclusions:

• The proposed method offers a computationally efficient alternative for profile likelihood approximation in PBPK modeling.
• This approach accelerates the assessment of parameter identifiability and confidence intervals, facilitating more robust PBPK model development.
• The technique holds promise for improving the speed and efficiency of PBPK model analysis and application.