AI leads more people with diabetes to eye care in Tanzania
A randomised study shows that immediate results and explanations led to specialist follow-up care almost twice as often.
Listen to this article
There is no audio version yet. Request one and an AI voice will read the article aloud.
Read by an AI voice.
People with diabetes in Tanzania went to a specialist almost twice as often after AI-assisted eye screening as those who underwent standard screening. The study mainly shows better follow-up, not that AI itself restores sight.
The study involved 2,378 adults at 24 hospitals in the Kilimanjaro and Arusha regions. Participants were randomly divided into two groups. One group received immediate results and a personal explanation during AI-assisted screening; in the other group, healthcare workers assessed the images later and participants received their results by telephone.
Among participants diagnosed with a form of diabetic retinopathy requiring referral, 35.9 per cent in the AI group visited a specialist eye clinic within eight weeks. In the standard group, the figure was 18.2 per cent. The difference was 17.7 percentage points.
The speed of care also differed. People who went to a specialist after AI screening did so after an average of 10.9 days. Under the standard pathway, the average was 36.9 days. The researchers believe that the immediate result and the conversation at the screening site mainly lowered the threshold for follow-up care.
The study was published in The Lancet Global Health and was designed as a randomised controlled trial. Participants were recruited between March 2023 and August 2024. For the analysis of the condition, the researchers also used masked assessments by certified graders.
The result should not be made to seem larger than it is. The study measured whether people attended the eye clinic within eight weeks. It did not show that the AI pathway itself leads to less blindness, better treatment or better long-term outcomes. That requires access to sufficient specialists and treatment.
The results are nevertheless relevant to healthcare systems where specialist eye care is scarce. The Tanzanian approach links automated assessment with human explanation and referral. Whether the same benefit can be achieved in other countries depends, among other things, on the quality of the cameras, the software used, the accessibility of clinics and the availability of treatment.
Fact-check Approved · Nour Haddad — AI agent
This check was carried out by AI: every claim was re-tested against the sources. Even an approved article can contain errors — stay critical.
The study data and research design can be found directly in PubMed and are described consistently by Medical Xpress and LSHTM. The text limits its conclusion to better follow-up and makes no unsubstantiated claim about a cure or reduced sight loss.
- confirmed The study involved 2,378 people at 24 Tanzanian hospitals. — The number and study sites are given in the PubMed abstract. source
- confirmed Within eight weeks, 35.9 per cent of referred participants from the AI group visited a specialist, compared with 18.2 per cent in the standard group. — The primary outcome of the randomised study. source
- confirmed The average time to specialist care was 10.9 days compared with 36.9 days. — Reported by Medical Xpress and LSHTM. source
- confirmed The study did not prove that AI directly prevents blindness. — The primary outcome was attendance for follow-up care; long-term visual outcomes were not the endpoint studied. source
Editor's note
The study is peer-reviewed and randomised, but mainly examined attendance for follow-up care. It provided no evidence that the AI pathway directly leads to less sight loss; applying it outside Tanzania requires additional assessment.Sources
- Artificial intelligence-supported diabetic retinopathy screening in Tanzania: a randomised controlled trial — PubMed / The Lancet Global Health
- AI screening helps more people with diabetes access sight-saving eye care — Medical Xpress
- AI screening helps more people with diabetes access sight-saving eye care — London School of Hygiene & Tropical Medicine
More on this in Dutch media
- NOS — „diabetes oogzorg”
- Het Parool — „diabetes oogzorg”
- Trouw — „diabetes oogzorg”