Google ARDA retinal AI tops 1M patients across three countries

Google's diabetic retinopathy screening model has crossed one million patients across hospitals in India, Thailand and Australia, according to a Comment published 23 September 2026 in Nature Medicine.

The paper is anchored by three deployment sites: the Aravind Eye Care System in Madurai, Rajavithi Hospital in Bangkok, and Lions Outback Vision at the Lions Eye Institute in Western Australia. Lead authors Richa Tiwari, Rajroshan Sawhney and Kasumi Widner at Google write with clinicians from each partner institution, including Paisan Ruamviboonsuk at Rajavithi Hospital and Angus Turner at the Lions Eye Institute.

The authors pitch the write-up as operational lessons rather than a model paper. "Scaling a successful deep learning tool across three highly distinct settings...offers cross-cutting insights that may inform the expansion of healthcare artificial intelligence globally," they write.

The tool at the center is ARDA, Google's Automated Retinal Disease Assessment system, which it first deployed in Madurai. Google's own account of the project allows that "the path to bringing medical AI into a real clinical environment was not easy." The retrieved abstract does not break out per-site sensitivity or specificity, and it does not disclose how screening costs or liability are apportioned between Google and the hospitals running the service. Two researchers we track flagged the paper to our radar this week.