AOP-Wiki EMOD 3.0: Data Model Expansions and Content Evaluation Framework for Using Agentic AI to Improve Integration between AOPs and New Approach Methodologies (NAMs)

Adverse Outcome Pathways (AOPs) are logic models that causally link measurable biological mechanisms to adverse outcomes relevant to chemical regulatory endpoints. They contextualize New Approach Methodologies (NAMs)—in vitro and in silico methods used as alternatives to animal testing—and the sequential events in an AOP serve as multi-scale models spanning biological scales. The AOP-Wiki is the global repository for AOPs and has played a central role in AOP expansion over the past decade. However, constraints within the current data model and application infrastructure limit the AOP-Wiki from supporting continued AOP growth and evolution. The transformative power of agentic AI has re-invigorated AOP-Wiki data modernization efforts at a time when core AOP principles can be harnessed to inform the use of AI for aggregating and structuring AOP-relevant information. Seizing upon this momentum, the authors present AOP-Wiki EMOD 3.0, the third in a series of evidence model prototypes, which concretely demonstrates data model expansions and a vision for how the AOP-Wiki might be transformed to better serve regulatory science and emergent use of AOPs in biomedical and One Health contexts. The work aims to lay a foundation to support computationally-generated AOPs and quantitative AOPs (qAOPs) by focusing on solutions for AOP-Wiki internal quality improvement, evidence structuring to enhance AOP FAIRness and AI-readiness, and improved integration between the AOP framework and NAMs to better serve next-generation risk assessment.

The authors developed AOP-Wiki EMOD 3.0 as a prototype evidence model that extends the existing AOP-Wiki data model. The methodology centers on three interconnected pillars: (1) internal quality improvement of the AOP-Wiki, (2) evidence structuring to enhance FAIRness (Findable, Accessible, Interoperable, Reusable) and AI-readiness, and (3) improved integration between AOPs and NAMs. The prototype concretely demonstrates data model expansions, likely including new entity types, relationships, and metadata fields to capture evidence provenance, quantitative relationships, and links to NAM data. The authors also propose a content evaluation framework for using agentic AI to aggregate and structure AOP-relevant information. While the abstract does not detail specific algorithms, the approach leverages agentic AI—autonomous AI agents capable of complex reasoning and task execution—to assist in populating, curating, and evaluating AOP content. The framework is designed to support computationally-generated AOPs and quantitative AOPs (qAOPs), which require structured, machine-readable data. The methodology emphasizes alignment with core AOP principles to guide AI use, ensuring that AI-generated content adheres to the causal logic and evidence standards of the AOP framework. The prototype serves as a proof-of-concept for transforming the AOP-Wiki into a more scalable, AI-ready platform.