Eye screening that reached a million people.
Doctors in India, Thailand and Australia describe what it actually took to run the same diabetic-eye-screening AI across more than a million patients — the practical lessons, not just the accuracy.
- Maturity
- Deployed, stage 4 of 4
- Support
- 1 source · paper
- Evidence detail
- How we know ↓
Diabetic retinopathy can be caught early with screening, but many people never get a timely eye exam.
A screening tool that works in one hospital does not automatically work across different health systems and countries.
A deep-learning tool read retinal images to flag people who needed a specialist referral, deployed across three health systems.
What was shown, and what wasn’t
Shown
A Nature Medicine Comment reports the tool was deployed across Aravind Eye Care System (India), Rajavithi Hospital (Thailand) and Lions Outback Vision (Australia), screening over one million patients, and draws cross-cutting lessons for scaling healthcare AI.
Not shown · limits
- This is a Comment reflecting on deployment, not a new controlled trial, and the work was funded by Alphabet Inc. with several authors employed by Alphabet; the lessons are the authors’ own account, not independent evaluation.
Bright editorial interpretation
It shows the human work behind scaling a medical AI — across clinics, countries and workflows — rather than a single accuracy figure.
Still open
How do screening yield and downstream treatment differ across these very different health systems?
Keep the evidence together · share a card
BRIGHT / EVIDENCE TO KEEP
Eye screening that reached a million people.
Possible benefit · Bright interpretation
It shows the human work behind scaling a medical AI — across clinics, countries and workflows — rather than a single accuracy figure.
Documented result · source-backed
A Nature Medicine Comment reports the tool was deployed across Aravind Eye Care System (India), Rajavithi Hospital (Thailand) and Lions Outback Vision (Australia), screening over one million patients, and draws cross-cutting lessons for scaling healthcare AI.
Limits & unknowns
This is a Comment reflecting on deployment, not a new controlled trial, and the work was funded by Alphabet Inc. with several authors employed by Alphabet; the lessons are the authors’ own account, not independent evaluation.
How do screening yield and downstream treatment differ across these very different health systems?
Evidence state: Deployed. Source check: 2026-09-27. A source check is not independent replication.
Practical lessons in the global scaling of clinical AI: from one hospital to over a million patients screened · Nature Medicine: https://www.nature.com/articles/s41591-026-04643-9
https://brightaifuture.com/discoveries/retinopathy-scaling
How we know1 source · checked 2026-09-27 · no corrections
Original sources
Institutions: Google · Aravind Eye Care System · Rajavithi Hospital · Lions Eye Institute
- Maturity
- Deployed
- Claim confidence
- medium
- Source published
- 2026-09-23
- Captured
- 2026-09-27
- Last source review
- 2026-09-27
- Editorial method
- AI-assisted source review
- Place / relevance
- India, Thailand and Australia · global-study
Bright compared this account with the linked original and supporting sources and kept reported, budgeted, projected, and observed claims distinct. Bright did not independently audit the underlying records.
Maturity describes the tested or operational setting. Confidence describes support for the particular claim; one does not determine the other.
Revision & correction history
2026-09-27 · Added to Bright from the 27 Sep 2026 intake as a bounded, source-checked record; company funding disclosed.
No corrections recorded.
Keep exploring
Explore the shared question in another setting. These connections do not imply replication.