NEWS AND VIEWS
18 February 2026 AI succeeds in diagnosing rare diseases An artificial-intelligence system uses clinical data, genetic information and literature searches to suggest diagnoses and provides the underlying reasoning. By Timo Lassmann ORCID: http://orcid.org/0000-0002-0138-2691 0 Timo Lassmann Timo Lassmann is at The Kids Research Institute Australia and the UWA Centre for Child Health Research, The University of Western Australia, Nedlands, Western Australia 6009, Australia. View author publications PubMed Google Scholar
For the estimated 300 million people1 living with a rare disease worldwide, obtaining an accurate diagnosis remains an arduous journey. The ‘diagnostic odyssey’ often takes five years or more2, and is marked by repeated specialist consultations, misdiagnoses and unnecessary treatments. With more than 7,000 rare diseases recognized3, most of which affect fewer than one in 2,000 individuals, even experienced clinicians struggle to connect unfamiliar patterns of symptoms with their underlying causes. Writing in Nature, Zhao et al.4 present an artificial-intelligence system called DeepRare, which uses a range of specialized tools and knowledge sources to generate ranked diagnostic hypotheses for rare diseases, each accompanied by reasoning that links the conclusions to verifiable medical evidence.
doi: https://doi.org/10.1038/d41586-026-00290-9
T.L. used Anthropic’s Claude Opus 4.5 to help prepare this News & Views article.
References Nguengang Wakap, S. et al. Eur. J. Hum. Genet. 28, 165–173 (2020). Faye, F. et al. Eur. J. Hum. Genet. 32, 1116–1126 (2024). Haendel, M. et al. Nature Rev. Drug Discovery 19, 77–78 (2019). Zhao, W. et al. Nature https://doi.org/10.1038/s41586-025-10097-9 (2026). Dawkins, H. J. S. et al. Clin. Transl Sci. 11, 11–20 (2018). Groza, T. et al. Hum. Genet. Genom. Adv. 7, 100558 (2026). Boiko, D. A., MacKnight, R., Kline, B. & Gomes, G. Nature 624, 570–578 (2023). Zou, J. & Topol, E. J. Lancet 405, 457 (2025). Download references
Competing Interests The author declares no competing interests.
Related Articles
Subjects