Generative models trained on electronic health records are viewed as ‘zero-shot predictors’ for clinical outcomes — but this interpretation is misleading.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$32.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 print issues and online access
$259.00 per year
only $21.58 per issue
Buy this article
- Purchase on SpringerLink
- Instant access to the full article PDF.
USD 39.95
Prices may be subject to local taxes which are calculated during checkout
References
Renc, P. et al. NPJ Digit. Med. 7, 256 (2024).
Waxler, S. et al. Preprint at https://doi.org/10.48550/arXiv.2508.12104 (2025).
Shmatko, A. et al. Nature 647, 248–256 (2025).
Pang, C. et al. Preprint at https://doi.org/10.48550/arXiv.2509.03643 (2025).
Bedi, S., Jain, S. S. & Shah, N. H. Nat. Med. 30, 2409–2410 (2024).
Falvey, A. Becker’s Payer Issues https://go.nature.com/4o0RW00 (1 July 2025).
Shah, N. H., Entwistle, D. & Pfeffer, M. A. JAMA 330, 866–869 (2023).
Brown, T. B. et al. In Advances in Neural Information Processing Systems 33, 877–1901 (NeurIPS, 2020).
Wornow, M. et al. NPJ Digit. Med. 6, 135 (2023).
Bedi, S., Jiang, Y., Chung, P., Koyejo, S. & Shah, N. JAMA Netw. Open. 8, e2526021 (2025).
Geirhos, R. et al. Nat. Mach. Intell. 2, 665–673 (2020).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Rights and permissions
About this article
Cite this article
Bedi, S., Fries, J.A. & Shah, N.H. How to interpret ‘zero-shot’ results from generative EHR models. Nat Med (2026). https://doi.org/10.1038/s41591-025-04094-8
Published:
Version of record:
DOI: https://doi.org/10.1038/s41591-025-04094-8
