Building the Foundations of AI-Driven Toxicology: How to Use Fragmented Data for Mechanism-Based Human Health Risk...

Area of Science:

• Environmental Health
• Toxicology
• Computational Biology

Background:

• Current toxicology databases are fragmented, lack standardization, and have imbalanced coverage, limiting their utility for regulatory decision-making.
• Shifting regulatory landscapes and technological advancements necessitate high-quality data for AI-based computational toxicology models.
• Existing data limitations hinder accurate prediction of environmental health risks.

Purpose of the Study:

• To outline challenges and propose strategic solutions for curating, harmonizing, and accessing toxicology data.
• To emphasize AI-enabled strategies for organizing and utilizing toxicological data infrastructures.
• To support the development of AI-driven environmental toxicology and mechanism-based human health risk assessment.

Main Methods:

• Developing confidence grading frameworks for heterogeneous data integration.
• Utilizing novel high-throughput platforms for accurate and efficient interaction data generation.
• Facilitating community-based data sharing and AI-enabled data infrastructure development.
• Integrating knowledge networks for mechanism-informed AI models and applying transfer learning.
• Leveraging knowledge graphs and prompt learning for predicting Exposure-Biology-Disease interactions.

Main Results:

• Proposed solutions address data curation, harmonization, and accessibility challenges.
• AI-enabled strategies enhance data organization, interoperability, and usability.
• Frameworks for integrating diverse data sources and building predictive models are presented.
• The development of systematic, interpretable, and predictive toxicological data systems is facilitated.

Conclusions:

• Building high-quality toxicology databases is crucial for advancing AI-driven toxicology.
• These databases provide a foundation for more reliable environmental health risk assessment.
• Enhanced data infrastructure will strengthen global environmental health protection through improved risk prediction and mechanistic understanding.