Deep learning marker of mortality risk from retinal photographs, validated in three external cohorts (RetiME)
A deep learning marker that maps a retinal photograph directly to an individual's mortality risk, trained on a European cohort, was associated with all-cause and cause-specific mortality and with the later onset of a wide range of systemic and eye diseases in three external cohorts in Southeast Asia, Eurasia and North America, and was used to reconstruct personalised lifespan trajectories.
- Model
- RetiME, a deep learning marker built on large foundation models (not named in the abstract)
- Field
- Medicine
- Date
- 2026-11-02
- Human collaborators
- Ting Xu, Ching-Yu Cheng
What was found
Retinal Mortality Equivalent (RetiME) was built on large foundation models and trained to read individual mortality risk from low-cost, non-invasive retinal photographs. Beyond a static risk estimate, the authors use it to reconstruct a personalised lifespan trajectory for each person, and report that these trajectories were associated with observed outcomes such as age at death. Clustering the trajectories separated cancer-predominant, cardiometabolic and neurodegenerative patterns. Interpretability analyses and a genome-wide analysis, which identified loci associated with RetiME, are offered as evidence that it captures biology rather than image artefacts.
Novelty check
Europe PMC, 2026-11-05: (TITLE:"retinal" OR TITLE:"fundus" OR TITLE:"retina") AND TITLE:"mortality" AND ("deep learning" OR "foundation model" OR "artificial intelligence") returned 14 records. Retinal deep learning markers of mortality are established, and two come from the same group: RetiAGE (Nusinovici and colleagues, Age and Ageing 2022) and a morbidity and mortality marker in Lancet Healthy Longevity (2024), both with Ching-Yu Cheng as senior author, alongside the retinal age gap of Zhu and colleagues (British Journal of Ophthalmology 2023). What is new here is the foundation-model base and the trajectory modelling, not the idea of reading mortality risk from the retina.
Caveats and known objections
Only the abstract and the PubMed record were read: the article redirected to a login, so the cohort names and sizes, the comparators, the effect sizes, the funding and competing-interest statements and whether code or weights are available are not recorded here. From the author list, the external cohorts were contributed by co-authors, including the US National Eye Institute and the Ufa Eye Research Institute in Russia, so this is external but not independent validation in this registry's sense, and the graded result is peer-reviewed. The design is retrospective, on cohorts collected for other purposes. Mortality is a hard outcome, but what is reported is association and prediction; no care decision used the marker and nothing shows that acting on it helps anyone. Autonomy is ai-assisted, the weaker reading: the model supplies the risk signal, but humans defined the target, the cohorts and every analysis, and no clinical use was tested.
No independent check is recorded on this entry yet.
Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.
Or on GitHub: submit a check challenge the grade send a correction or send a pull request
Flag this for triage
Signals order the review queue and nothing else. They are never published, and they never move a grade: that takes a citation.
Entry history (1 event)
- AddedEntered the registry graded Peer reviewed and AI-assisted.
Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.
Graded peer reviewed for verification and ai-assisted for autonomy. What these mean.
Cite this entry
whataifound.org. (2026). Deep learning marker of mortality risk from retinal photographs, validated in three external cohorts (RetiME). whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2026-11-02-retime-retinal-mortalityBibTeX
@misc{whataifound-nationaluniversityofsingaporeandsingaporeeyeresearchinstitute-2026-mortality,
title = {Deep learning marker of mortality risk from retinal photographs, validated in three external cohorts (RetiME)},
author = {{whataifound.org}},
year = {2026},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by National University of Singapore and Singapore Eye Research Institute. Verification: Peer reviewed. Autonomy: AI-assisted.},
url = {https://whataifound.org/finding/2026-11-02-retime-retinal-mortality}
}