Model verification tools: a computational framework for verification assessment of mechanistic agent-based models.

Area of Science:

• Computational Biology
• Regulatory Science
• Medical Device Standards

Background:

• Computer modeling and simulation are increasingly vital in life sciences for assessing medicinal product safety and efficacy.
• Mechanistic Agent-Based Models (ABMs) are gaining traction for in silico trials, but lack standardized verification methods for regulatory approval.
• The VV&UQ (Verification, Validation, and Uncertainty Quantification) standard, designed for medical devices, offers a adaptable framework for ABM verification.

Purpose of the Study:

• To propose a set of automated tools for the verification assessment of mechanistic Agent-Based Models.
• To adapt and apply the VV&UQ standard's principles to the verification of in silico trials for medicinal products.

Main Methods:

• Development of an automated computational workflow for ABM verification.
• Application of the verification framework to a specific Agent-Based Model used in a COVID-19 in silico trial.

Main Results:

• A functional verification framework for mechanistic Agent-Based Models has been developed.
• The framework was successfully applied to a COVID-19 Agent-Based Model simulation, demonstrating its practical utility.

Conclusions:

• The proposed verification computational workflow facilitates robust and correct Agent-Based Models.
• This approach provides strong evidence necessary for meeting regulatory requirements for in silico trials.