Optimal Experimental Design Based on Two-Dimensional Likelihood Profiles.

Area of Science:

• Systems Biology
• Computational Biology
• Mathematical Modeling

Background:

• Dynamic biological systems are often modeled using non-linear ordinary differential equations.
• Estimating model parameters and their uncertainty is crucial for interpreting results.
• High parameter uncertainty hinders biological model interpretation and requires efficient experimental design.

Purpose of the Study:

• To develop a novel experimental design method for non-linear models that minimizes parameter uncertainty.
• To address limitations of existing Bayesian and frequentist approaches in handling non-linearity and prior distributions.
• To provide a design criterion that quantifies expected parameter uncertainty for specific experimental conditions.

Main Methods:

• Proposed a two-dimensional profile likelihood approach for experimental design.
• Implemented the method within the open-source Data2Dynamics toolbox in Matlab.
• Demonstrated applicability using a systems biology model with censored data to simulate parameter uncertainty.

Main Results:

• The two-dimensional profile likelihood approach provides a meaningful criterion for experimental design.
• The method successfully identified optimal experimental conditions to reduce parameter uncertainty.
• Validation on a systems biology model confirmed the approach's feasibility for real-world applications.

Conclusions:

• The developed method enhances the efficiency of parameter estimation in non-linear dynamic systems.
• It offers a robust frequentist alternative for experimental design in systems biology.
• The approach is broadly applicable to non-linear modeling tasks beyond systems biology.