Optimal experiment design for practical parameter identifiability and model discrimination

Area of Science:

• Systems Biology
• Mathematical Biology
• Computational Biology

Background:

• Mechanistic biological models require parameter estimation for validation and prediction.
• Model identifiability, the confidence in parameter determination from data, is crucial.
• Experimental design significantly impacts data informativeness for parameter inference.

Purpose of the Study:

• To develop methods for optimal experimental design to maximize parameter identifiability.
• To optimize control inputs for experiments to enhance parameter estimation.
• To improve model discrimination between competing biological models.

Main Methods:

• Utilized a profile likelihood approach to assess parameter identifiability.
• Formulated optimal experimental design for model discrimination as an optimal control problem.
• Applied Pontryagin's Maximum Principle for efficient problem-solving.

Main Results:

• Demonstrated techniques for optimal control design in experiments.
• Showcased the application in ordinary differential equation models.
• Enhanced the ability to distinguish between different biological models.

Conclusions:

• Optimal experimental design, particularly control input optimization, is key for robust biological model validation.
• The presented optimal control framework effectively addresses parameter identifiability and model discrimination challenges.
• This approach provides a powerful tool for advancing quantitative biological research.