Challenges and opportunities for validation of AI-based new approach methods

Area of Science:

• Toxicology and Computational Science
• Chemical Safety Assessment
• Regulatory Science

Background:

• Artificial intelligence (AI) presents a paradigm shift in toxicology, offering potential to streamline new approach methods (NAMs) validation.
• Traditional validation methods face limitations, necessitating innovative approaches for chemical safety assessment.
• The integration of AI in NAMs promises enhanced predictive power and efficient data utilization.

Purpose of the Study:

• To explore the challenges, opportunities, and future directions for validating AI-based NAMs in toxicology.
• To present AI-powered frameworks, such as e-validation, for overcoming traditional validation limitations.
• To propose robust validation strategies and discuss ethical considerations for AI in toxicology.

Main Methods:

• Review of current literature on AI in NAMs validation.
• Conceptualization of an AI-powered framework (e-validation) for streamlining validation processes.
• Proposal of validation strategies including tiered approaches, benchmarking, and uncertainty quantification.

Main Results:

• AI-based NAMs offer enhanced predictive toxicology and reduced reliance on animal testing.
• Key challenges include data quality, model interpretability, and regulatory acceptance.
• Opportunities lie in efficient data integration and improved mechanistic understanding.

Conclusions:

• AI-driven NAMs hold transformative potential for chemical safety assessment.
• Robust validation, ethical considerations, and collaborative efforts are crucial for successful AI integration.
• Companion AI agents can ensure ongoing method validity and adaptation to evolving AI developments.