Employing Observability Rank Conditions for Taking into Account Experimental Information a priori

Area of Science:

• Systems Biology
• Control Theory
• Mathematical Modeling

Background:

• Model identifiability is crucial for parameter inference from dynamic systems.
• Distinction between structural identifiability (model-based) and practical identifiability (data-influenced).
• Observability analysis offers a framework for studying structural local identifiability.

Purpose of the Study:

• To investigate the utility of observability-based methods for assessing practical identifiability.
• To explore extensions of rank tests for informing experimental setup.
• To determine the scope and limitations of these methods for practical applications.

Main Methods:

• Utilizing generalized observability matrix and rank computations.
• Developing and exploring extensions of rank tests.
• Comparing symbolic (a priori) and numerical (a posteriori) identifiability analyses.

Main Results:

• Observability-based methods can offer insights into practical identifiability beyond structural analysis.
• Extensions of rank tests show potential for guiding experimental design.
• The effectiveness of these extensions varies depending on the specific model and data characteristics.

Conclusions:

• Observability analysis provides a valuable, unified approach to studying both structural and practical identifiability.
• Further development of rank test extensions is needed to fully leverage their potential for experimental design.
• The study highlights the interplay between model structure, data, and experimental design in achieving practical identifiability.