A Bayesian-based Screening Framework for Optimal Development of Safe-by-design Nanomaterials

In this article we propose a novel screening approach for the development of new nanomaterials (NM) which are safe(r)-by-design. An integral part of our approach involves the use of recently proposed notation standards and extensions, which facilitate the representation and processing of data related to NMs. In particular, we utilize a general extendable representation for NMs which is built in line with InChI extensions for NMs. This representation enables us to link the composition, structure, and experimental conditions of a NM to its toxicity profile, using descriptor information obtained from available datasets. The processed data are then complemented with predictions from QSAR models and fed to a Bayesian-based procedure to assess the safety of new NM designs. This comprehensive approach can be used to build a local screening tool to search for less hazardous NM configurations.