Friday, 10 May 2024 14:20

Transformers: fewer toxicity tests on animals thanks to deep learning Featured

Swedish researchers from Chalmers University of Technology and Gothenburg University have developed an AI method to better identify toxic chemicals based on molecular structure.

According to a report, this would make it possible to better control the ever-growing number of chemicals used in society and reduce the number of animal experiments. The method is based on Transformers, an AI model for deep learning that was originally developed for language processing.

The scientific team investigated the AI in the field of aquatic toxicity testing. The model showed high predictive performance for all common groups of organisms tested - algae, aquatic invertebrates and fish. It also has an advantage over the QSAR methods that are already frequently used, as the scientists write. The applicability domain is larger and there are significantly fewer errors.

Original publication:
Gustavsson M, Käll S, Svedberg P, Inda-Diaz JS, Molander S, Coria J, Backhaus T, Kristiansson E. Transformers enable accurate prediction of acute and chronic chemical toxicity in aquatic organisms. Sci Adv. 2024 Mar 8;10(10):eadk6669. doi: 10.1126/sciadv.adk6669. Epub 2024 Mar 6. PMID: 38446886; PMCID: PMC10917336.

Further information: