Moseley Publishes Invited Commentary in Nature
LEXINGTON, Ky. (Feb. 26, 2024) - Hunter Moseley, PhD, professor in the UK College of Medicine and co-director of biomedical informatics for the UK Center for Clinical and Translational Science, recently published an invited commentary in the prestigious journal Nature.
His article, titled "In the AI science boom, beware: your results are only as good as your data", argues that achieving trustworthy results through machine-learning systems requires intensive vetting of data both before and after publication.
Moseley and his lab reviewed state-of-the-art machine-learning methods for predicting the metabolic pathways that metabolites belong to, on the basis of the molecules’ chemical structures. They aimed to find, implement and potentially improve the best methods for identifying how metabolic pathways are perturbed under different conditions: for instance, in diseased versus normal tissues.
The team found several papers, published between 2011 and 2022, that demonstrated the application of different machine-learning methods to a gold-standard metabolite data set derived from the Kyoto Encyclopedia of Genes and Genomes (KEGG), which is maintained at Kyoto University in Japan. Their review suggested that the algorithms improved over time, with newer methods performing better than older ones. But were those improvements real?
