Data-Driven Model Validation Across Dimensions.

Area of Science:

• Mathematical and computational biology
• Biophysics
• Systems biology

Background:

• Dimensionality reduction (e.g., 3D to 2D) is common in computational biology models.
• Experimental data (e.g., flow cytometry) often represents 3D biological systems.
• Comparing models and data across dimensions is crucial for validation.

Purpose of the Study:

• To develop a methodology for scaling 2D model outputs to 3D experimental data.
• To enable accurate model calibration and validation across different dimensions.
• To address the challenge of comparing reduced-dimension models with full-dimension experimental datasets.

Main Methods:

• Proposing a novel scaling methodology.
• Applying the method to agent-based models of granuloma formation.
• Applying the method to agent-based models of skeletal muscle tissue.
• Evaluating method accuracy using simulated data.

Main Results:

• Demonstrated a technique to bridge the dimensional gap between models and data.
• Successfully applied the scaling method to biological examples.
• Quantified the accuracy of the proposed methodology in controlled scenarios.

Conclusions:

• The developed methodology facilitates robust data-driven model validation in computational biology.
• This approach enhances the utility of reduced-dimension models by enabling comparison with 3D experimental data.
• The technique is applicable to diverse biological systems and modeling paradigms.