Bright key facts / Deployed
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.
- AI’s role
- A deep-learning tool read retinal images to flag people who needed a specialist referral, deployed across three health systems.
- Documented result
- 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.
- Important limitation
- 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.