Researchers from Friedrich Schiller University Jena, Helmholtz-Zentrum Berlin, HIPOLE Jena, and the Swiss software company Zakodium have developed an artificial intelligence system called SECS that supports the analysis of chemical structures from spectroscopic data.
The AI combines machine learning with evolutionary algorithms to interpret spectroscopic measurements and propose plausible molecular structures. In benchmark tests, SECS correctly identified the target structure as its top prediction in more than 80% of cases and achieved performance comparable to experienced chemists.
Rather than replacing human expertise, the system serves as a valuable tool for supporting structure elucidation, helping researchers validate results and identify potential inconsistencies in their analyses. The source code, model data, and a public test version of SECS are openly available to the scientific community.
Further information is available on the University of Jena website: https://www.uni-jena.de/en/427152/kuenstliche-intelligenz-wertet-chemische-spektren-in-minuten-aus