Optimal initial state for fast parameter estimation in nonlinear dynamical systems

Area of Science:

• Dynamical Systems Modeling
• Computational Mathematics
• Pharmacokinetics

Background:

• Parameter estimation in dynamical models is crucial but challenging, especially under bounded error conditions.
• Existing methods often face limitations in precision and computational time.

Purpose of the Study:

• To enhance parameter estimation precision and reduce computational time for dynamical models.
• To address challenges in bounded error contexts using an optimal initial state design.

Main Methods:

• An optimal initial state design combined with a volumetric criterion-based contractor.
• Sensitivity analysis to identify optimal initial states.
• Elementary effect analysis for efficient sampling time selection.
• Guaranteed state and sensitivity estimation under interval uncertainty.

Main Results:

• The proposed methodology was successfully applied to a pharmacokinetics model.
• Simulation results demonstrated significant improvements in parameter estimation.

Conclusions:

• The developed methodology offers a substantial improvement in both computational time and precision for parameter estimation.
• The approach shows strong potential for application in complex dynamical systems, particularly in pharmacokinetics.