Google ARDA eye AI passes million patients across three countries

Google says a single deep-learning eye-screening tool has now been used on more than a million patients across India, Thailand and Australia. That is the headline figure of a Comment published in Nature Medicine on September 23, 2026, drawing lessons from three anchor deployment sites.

Those sites are the Aravind Eye Care System in Madurai, Rajavithi Hospital in Bangkok, and Lions Outback Vision in Western Australia. The lead authors are Richa Tiwari, Rajroshan Sawhney and Kasumi Widner from Google, writing with clinicians at each partner institution. The piece frames its own scope: scaling a deep-learning tool "across three highly distinct settings" that "offers cross-cutting insights that may inform the expansion of healthcare artificial intelligence globally."

The underlying tool is ARDA, Google's Automated Retinal Disease Assessment. In an earlier post the company said the model had "supported more than 600,000 screenings in clinics around the world," and that its licensed partners Forus Health, AuroLab and Perceptra are targeting "a combined 6 million AI-supported screenings to resource-constrained communities at no cost to patients" across India and Thailand over the next decade, working alongside Thailand's Ministry of Public Health.

Two of the researchers we track flagged the paper the day it published. As a Comment rather than a trial writeup, it does not publish per-country accuracy figures, camera-model breakdowns or false-referral rates; the framing is workflow and deployment lessons rather than clinical performance numbers.