AI-assistance in Knowledge Collection and Curation supporting Safe and Sustainable by Design workflows
We discuss the use of data science, modelling and AI techniques to assist the collection, generation and use of evidence as a critical strategic New Approach Method (NAM) methodology supporting Safe and Sustainable by Design (SSbD) goals. Valuable assistance is obtained from transformative tools that use data and scientific techniques to perform more accurate, ethical, and comprehensive assessments, embedded into SSbD decision-making workflows. To achieve this goal, full use of existing data, models and knowledge should be leveraged into providing relevant information supporting assessment and decision goals between alternatives in early stage innovation. Such data needs to be integrated with evidence-weighting and scoring schemes, including uncertainty, to reach initial decisions on alternatives and to plan for subsequent refinement phases. The findings may include recommendations for generating the most meaningful and useful data to address gaps and uncertainty. We demonstrate the methodology followed and questions arising from current SSBD4CheM case study work applying SSbD to chemicals, polymers and advanced materials. We present on resources we have developed for: Collection, extraction and curation of knowledge from literature, toxicology databases and safety data sheets; Demonstration and access to knowledge collection tools and resources; Hazard profiling of product formulations and ingredients; Knowledge infrastructure for characterisation of materials; SSbD Workflows documenting case study results.
AI-assistance in Knowledge Collection and Curation supporting Safe and Sustainable by Design workflows
- 1. This project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement n° 101138475. UK participants in SSbD4CheM project are supported by UKRI. CH participants in SSbD4CheM project receive funding from the Swiss State Secretariat for Education, Research and Innovation (SERI). AI-assistance in Knowledge Collection and Curation supporting Safe and Sustainable by Design Workflows Barry Hardy (Edelweiss Connect) SSbD4CheM Stakeholder Outreach Webinar: Practical Application and Case Studies in Industrial SSbD; 11 June 2025
- 2. Knowledge Infrastructure and Workflow approach
- 3. Tier 0/1 workflow for SSbD
- 4. Tier 0/1 workflow for SSbD (cont)
- 5. Step 1 Hazard Characterisation Used tools : JANUS OECD QSAR TB VEGA STOPTOX SAFERSKIN ZZS SIMILARITY Used databases : PUBCHEM ECHA HAZ-MAP COMPTOX CAMEO CHEMICALS EC, JRC; 2022. Available from: doi/10.2760/487955
- 6. From databases From in silico tools Step 1 Hazard Characterisation
- 7. Continue SSbD Assessment Step 1 Hazard Characterisation
- 8. Current Workflow for Polymers suggested by REACH Grouping of polymers TIER 1: Non-vertebrate animal base set TIER 2: repeated exp., systematic toxicity set TIER 3: further investigation of end points of concern Chemistry and Physicochemical properties Bioavailability Environmental and Human Health Hazard Data RESULT Potential risk or insufficient data Focused testing Targeted ecotoxicity test In vivo studies Systemic toxicity test Detailed testing Environmental simulation test Comprehensive in vivo studies Potential risk or insufficient data RISK No RISK Modera teRISK ● in vitro skin and eye irritation/ corrosion ● in chemico/in vitro skin sensitization and in vitro genotoxicity ● in vitro local effects on respiratory epithelia, gastrointestinal epithelia ● systemic bioavailability assessment ● environmental fate and distribution ● aquatic environment, acute ● terrestrial environment, acute ● bioavailability
- 9. Characterisation Workflow (based on ACCORDs)
- 10. Analysis, Editing etc. Easy figure creation for publication Data and metadata overview Organised folder structure Workflow for Image Analysis
- 11. ● Application Software and Database Modules for Sub-Ingredients, Raw Ingredients, Contaminants and Formulas ● Extraction of INCI profile documents for Cosmetic Ingredients to create a reference Sub-Ingredient Database for SmartSafety ● Formula Compare (identification of most similar formulas) and Margin of Safety calculations ● Generate CPSR, CSAR and other reports (with variations) ● Deployment onto cloud infrastructure SmartSafety: Formulation and Ingredient Risk Assessment and SSbD modules
- 12. SaferSkin Integration with SmartSafety supporting NAMs-based Skin Safety Assessment of Ingredients in Formulation
- 13. Reference JK Saxe, L Hoffman, R Labib, Method to incorporate green chemistry principles in early-stage product design for sustainability: case studies with p ersonal care products , Green Chemistry, 2022, 24, 4969–4980. DOI: https://doi.org/10.1039/D2GC00842D SmartSafety – Environmental Scoring of Ingredients in Formulation
- 14. Automated AI-assisted Safety Data Sheet Retrieval AI https://sds-finder.edelweissconnect.com/
- 15. 15 Biological Data (Biostudies, EdelweissData, Notebooks) Knowledge Graphs Compound Database Methods Database (ToxTemp) Prediction and Analysis Tools (QSAR, Molecular Simulations, Metabolism, Omics) Toxicokinetics and Exposure protocols OpenModel Repository Case Study Documentation AI assisted AI assisted AI assisted AI assisted Life Cycle Assessment Software for documenting workflows
- 16. Demonstration of Tools and Resources supporting SSbD https://ssbd4chem.edelweissconnect.com/welcome
- 17. Demonstration – Test Method Documentation (ToxTemp Reference Database) https://risk-hunt3r.net/
- 18. Demonstration – Workflow Documentation of NAMs-based NGRA and SSbD
- 19. Demonstration – Materials Characterisation Knowledge Infrastructure
- 20. Demonstration – Automated Collection and Extraction of Safety Data https://sds-toolbox.edelweissconnect.com/
- 21. Demonstration – Automated Collection and Extraction of Literature Data https://pubmed-cheminsight.edelweissconnect.com/
- 22. Demonstration – Automated Collection and Extraction of Chemcial Data https://pubchempal.edelweissconnect.com/
- 23. Demonstration – Automated Collection and Extraction of Regulatory Data https://observablehq.com/@tomazm/chemical-similarity-search
- 24. Demonstration – chemical similarity search for regulatory toxicology data
- 25. Characterization Human health and environmental safety assessment Field-Flow Fractionation hyphenated to light scattering Exposure assessment / risk management Sustainability assessment across the life cycle assessment Computer aided (re)design approach High resolution chemical imaging Reduced TVOC emissions Zebrafish embryo model Machine learning toxicity models LCA, s-LCA, and LCC Computational Fluids Dynamics Time-of-flight mass spectrometry Molecular simulations hiPSC-based in vitro lung model Human skin ex vivo and in vitro models Characterisation of microfibers and microfibrils release SSbD4CheM framework and workflow Fate and the behaviour in the environment D&C&E LCA, s-LCA, and LCC LCA, s-LCA, and LCC SSbD4CheM Framework and Case Studies
- 26. This project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement n° 101138475. UK participants in SSbD4CheM project are supported by UKRI . CH participants in SSbD4CheM project receive funding from the Swiss State Secretariat for Education, Research and Innovation (SERI). Project partners Funding acknowledgements