Author Correction: Learning the natural history of human disease with generative transformers

"Author Correction: Learning the natural history of human disease with generative transformers Author Correction Open access Published: 12 November 2025 Artem Shmatko orcid.org/0000-0002-2184-08561,2,3 na1, Alexander Wolfgang Jung orcid.org/0000-0001-8416-68492,4,5,6 na1, Kumar Gaurav2 na1, Sren Brunak orcid.org/0000-0003-0316-58664,7, Laust Hvas Mortensen5,7,8, Ewan Birney orcid.org/0000-0001-8314-84972, Tom Fitzgerald orcid.org/0000-0002-2370-84962 & Moritz Gerstung orcid.org/0000-0001-6709-963X1,2,9,10,11,12 Nature (2025)Cite this article Computational science Diseases Risk factors Correction to: Nature https://doi.org/10.1038/s41586-025-09529-3 Published online 17 September 2025"

"In the version of this article initially published, in the Methods Exponential waiting time model section, there was a typographical error in the equation now reading lossj = log P(j) = cross_entropy(logits, tokens), where a minus sign originally appeared before the cross-entropy term. The equation has been amended in the HTML and PDF versions of the article."

A typographical error placed an incorrect minus sign before the cross-entropy term in an exponential waiting time model loss equation. The corrected expression for the loss is loss_j = log P(j) = cross_entropy(logits, tokens). The incorrect sign would invert the intended loss contribution and could affect training dynamics or interpretation of log-probabilities. HTML and PDF versions have been updated to reflect the corrected equation. The correction was issued and published on 12 November 2025.

Read at www.nature.com