AI-assisted Next Generation Risk Assessment and Safe and Sustainable by Design Workflows enabled by FAIR Data and Knowledge

We describe our work in preparing datasets, models and knowledge graphs according to FAIR (Findable, Accessible, Interoperable, Reusable) principles connecting compound information, biological response data, pathways and key events based on our experiences in developing knowledge resources supporting Next Generation Risk Assessment (NGRA) and Safe and Sustainable by Design (SSbD) case studies, applications and workflows. Example knowledge resources to be discussed include:
a) FAIR datasets, models and knowledge graphs supporting New Approach Methods in Toxicology;
b) integration of knowledge in workflows supported tiered testing and assessment strategies for NGRA problem formulations and case studies;
c) integration of evidence for the assessment of ingredients and formulations using tiered SSbD workflows.

The related open knowledge resources include harmonised data templates for organising research data according to FAIR principles; omics data management best practices and implementation; approach to collected data including harmonisation, referencing and integrity goals; compound data management including approach to compound identifiers, properties, project and case study annotation, characterisation and linked biological and toxicological data, biological models and support for user-oriented hypothesis testing. 

Such resources can be used in the support of AI-enabled NGRA and SSbD case study work and its effective documentation including: (a) Implementing FAIR data principles; (b) Enabling effective retrieval of reliable well-referenced linked background knowledge organised into a knowledge graph; (c) Enhanced toxicology learning of Machine Learning- and Generative AI- based models; (d) Incorporation of AI knowledge into risk assessment and SSbD workflows and tasks, including analysis, evidence interpretation and decisions; (e) Generation of reports documenting workflow processes, including high integrity auditable supporting data and knowledge records; (f) Incorporating trust, interpretability and transparency; (g) Alignment with emerging AI guidance and legislation.

The presentation includes work carried out on several EU projects including EU-ToxRisk, RISK-HUNT3R, ASPIS, ACCORDs, SSbD4CheM and BioPhenom.