Can a same-visit AI eye exam close a screening gap for young people with diabetes?
In a prospective US study of 397 young people, point-of-care autonomous AI brought exam completion to 99.5% among White participants and 98% among non-White participants. That closes a process gap—not yet a vision-outcome gap.
The short version
- What happened
- The prospective ACCESS2 study enrolled 397 young people: 72% had type 1 diabetes and 28% had type 2 diabetes.
- Who it affects
- Young people and their families may be able to complete a recommended retinal screen during the diabetes appointment instead of organising a separate visit. Health systems would still need reliable referral, follow-up and quality assurance after the AI result.
- What remains unproven
- This was a prospective, single-workflow study rather than a randomised comparison with another same-visit screening approach.
Keep in mind
This was a prospective, single-workflow study rather than a randomised comparison with another same-visit screening approach.
The sensitivity estimate has a wide 95% confidence interval of 67.6% to 100%, consistent with a small number of positive cases.
Completion does not establish that the programme prevents sight loss, improves long-term follow-up or works similarly in other health systems.
Digital Diagnostics supplied AI interpretations in kind; one author reports company roles, investment and relevant patents.
Source analysis · 1 linked source. Primary analysed source date: 23 October 2026.
Keep following the evidence
Join The Impact Brief founding list for selected AI reports, their effect on people and the evidence limits. Read a sample first.
Read the full report · 3 min The evidence, methods, implications and limitations
Putting the exam inside diabetes care almost eliminated the completion gap
ACCESS2 recruited 397 White and non-White young people with diabetes during routine diabetes-care visits. Each participant received an autonomous-AI retinal exam and optical-coherence-tomography imaging at the point of care. Before enrolment, prior eye-exam completion differed sharply: 79.8% among White participants versus 61.8% among non-White participants. With the study workflow, 99.5% and 98% completed the AI exam, meeting the researchers' equivalence test.[1]
Completion was strong; diagnostic certainty was less complete
Compared with the study's standard grading for more-than-mild diabetic retinopathy, the AI system reported 100% sensitivity, but the 95% confidence interval ran from 67.6% to 100%, signalling few positive cases and substantial uncertainty. Specificity was 76.0% (95% CI 71.4% to 79.9%). A completed test is valuable only if its referrals and reassurances are dependable, so both accuracy and downstream follow-up matter.[1]
The distinction that matters
What ACCESS2 measured
The study links an access question to a diagnostic-accuracy check, but stops before patient outcomes.
Before the study
Check prior eye-exam completion
79.8% White; 61.8% non-White
Same-visit AI
Image and interpret during diabetes care
99.5% White; 98% non-White completed
Accuracy check
Compare AI with standard grading
100% sensitivity; 76% specificity, with uncertainty
This is evidence about access, not prevented sight loss
For families, a same-visit result can remove the need to arrange a separate specialist screening appointment. But the study did not compare AI with a same-visit human examiner, randomise participants to different workflows or measure later visual outcomes. It therefore cannot separate the effect of autonomous interpretation from the effect of bringing retinal imaging into the diabetes clinic. Digital Diagnostics supplied interpretations in kind, and a co-author is an investor, director and consultant at the company with relevant patents. Replication across health systems, transparent false-positive follow-up and longer-term outcomes would strengthen the case.[1]
Watch the background, then check the evidence
Illustrative photo · not a video frame
The Royal Society · 44 min 28 sec · Background explainer
Is this the AI future we want? | The Royal Society
YouTube loads only when you choose; playback starts when you press play.
A Royal Society discussion with Madhumita Murgia and biomedical engineer Alison Noble on AI’s potential in healthcare, science and culture, recorded at the 30 June 2026 event.
Evidence check: A discussion of opportunities and risks, including future possibilities. It is not evidence that every application described is available or clinically proven.
Published 2026-07-01. Source checked 14 October 2026. Check captions on the publisher’s video page.
Check the healthcare evidencePhoto: Daniel Sone / National Cancer Institute / Unsplash · Licence
Continue the story
Read next
Continue exploring paediatric diabetes: compare the finding and its evidence limits.
Can an AI agent predict diabetes outcomes in children? A 6,507-patient preprint is promising—not clinical proof
The evidence-citing system ranked future outcomes better than the tested statistical models on average. Its advantage varied by outcome, came from one hospital and has not been prospectively tested in care.
Read the next reportExplore the full source and evidence record
What this means for people
- Young people and their families may be able to complete a recommended retinal screen during the diabetes appointment instead of organising a separate visit. Health systems would still need reliable referral, follow-up and quality assurance after the AI result.
Global context
Screening gaps often reflect appointment burden, transport, specialist availability and fragmented care. The transferable idea may be point-of-care delivery; other countries would need locally validated accuracy, referral capacity, reimbursement and governance before adopting the same model.
What the evidence does not yet show
- This was a prospective, single-workflow study rather than a randomised comparison with another same-visit screening approach.
- The sensitivity estimate has a wide 95% confidence interval of 67.6% to 100%, consistent with a small number of positive cases.
- Completion does not establish that the programme prevents sight loss, improves long-term follow-up or works similarly in other health systems.
- Digital Diagnostics supplied AI interpretations in kind; one author reports company roles, investment and relevant patents.
What to watch next
- The number and outcomes of false-positive referrals
- Replication outside one US clinical setting
- Long-term specialist follow-up and visual outcomes
- Head-to-head comparisons with other same-visit screening workflows
Living evidence record
Impact record IAI-1T4SEIV
Evidence stage
Studied
Confidence
Supported
Reporting basis
Source analysis
Independent or research support
Not yet
Record status
Monitoring
Last checked
23 October 2026
Source trail
1 direct source across 1 source type.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Single-source reporting disclosure
This record analyses one direct source. It can establish what Nature Communications published or reported, but it is not independent corroboration of every performance claim or predicted outcome. The confidence label will change only when broader evidence is added.
Evidence trail
Sources used for this report
Links checked 23 October 2026
Research question, topics and regions
Autonomous diabetic-retinopathy screening in young people
Topics: Diabetic retinopathy · Point-of-care screening · Health equity · Autonomous AI · Paediatric diabetes
Regions: United States · North America
This report is labelled source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
Reader commentary
What does this mean where you are?
Tell us what you have seen at work, in your community or in everyday life. A question is welcome too. No account is needed; we review notes before publishing them.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.