Artificial Intelligence-Assisted Safe and Sustainable by Design Workflows applied to Case Studies on Chemicals and Materials

We describe the Artificial Intelligence (AI) and data science methodology and tools we have used and the results we have obtained in case studies based on the Safe and Sustainable by Design (SSbD) framework proposed by the European Joint Research Centre. 

Our case study application includes applying SSbD to 
a) cosmetic formulations including micro and nano polymeric forms, 
b) alternative PFAS-free thinner coatings for textiles, 
c) consideration of biopolymers as alternatives to current polymers used in products such as automobiles, 
d) use of bio-sourced chemicals as alternatives to carbon in fibres and composites, and e) graphene materials.  

Valuable assistance in the risk identification and characterisation work is obtained from 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 is leveraged by use of AI assistance into providing relevant information supporting assessment and decision goals between alternatives in early stage innovation. Such data is integrated with evidence-weighting and scoring schemes, including uncertainty, to reach initial decisions on alternatives and to plan for subsequent refinement phases. The findings include recommendations for generating the most meaningful and useful data to address gaps and uncertainty. We demonstrate the methodology and tools followed and questions arising from current case study work applying SSbD to chemicals, polymers and advanced materials including:

* Using AI-assisted resources to find, extract and curate knowledge from public databases and literature including resources we are sharing openly with the community;
* Integration of SSbD into early stage risk assessment decision making on ingredients and formulations;
* Hazard profiling and comparison of product formulations and ingredients based on existing knowledge;
* Knowledge infrastructure for characterisation of materials including AI-assisted labelled image data analysis of experimental microscopy results;
* Use of biological data from New Approach Methods for enriching hazard and exposure characterisation;
* SSbD Workflows documenting case study results.