Advances in Surrogate Modeling for Biological Agent-Based Simulations: Trends, Challenges, and Future Prospects

Area of Science:

• Computational Biology
• Biomedical Informatics
• Systems Biology

Background:

• Agent-based modeling (ABM) is crucial for understanding complex biological and biomedical systems.
• High computational demands limit ABM's widespread application, especially with increasing model complexity and the curse of dimensionality.
• Parameter exploration and uncertainty analysis are often computationally prohibitive in traditional ABMs.

Purpose of the Study:

• To review traditional and surrogate-assisted methodologies for enhancing ABM computational efficiency.
• To synthesize recent advancements in surrogate-assisted approaches for biological and biomedical applications.
• To identify challenges and future research directions for surrogate-assisted ABMs.

Main Methods:

• Examination of traditional methods for parameter estimation, sensitivity analysis, and uncertainty quantification within ABMs.
• Synthesis of statistical, mechanistic, and machine learning-based surrogate modeling techniques.
• Emphasis on hybrid strategies combining mechanistic insights with machine learning.

Main Results:

• Surrogate modeling significantly reduces computational runtime for ABM analysis.
• Various surrogate approaches (statistical, mechanistic, ML, hybrid) offer different balances of interpretability and scalability.
• Emerging hybrid strategies show promise for integrating domain knowledge with data-driven methods.

Conclusions:

• Surrogate-assisted modeling is a viable solution to the computational limitations of ABMs.
• Hybrid approaches offer a promising path for interpretable and scalable ABM analysis.
• Standardized benchmarks are needed to advance methodological rigor and adoption in biology and medicine.