261-OR: Autonomous Artificial Intelligence (AI) Testing for Diabetic Eye Disease (DED) Closes Care Gap and Improves Health Equity on a Systems Level
Abstract
We aimed to examine the change in adherence to annual diabetic eye disease (DED) testing in an integrated healthcare system (Johns Hopkins Community Physicians 30+ community based primary care clinics) before and after deployment of autonomous artificial intelligence (AI). Adherence to annual DED testing is defined as completed evaluation by either a human ophthalmology provider, or autonomous AI within a given calendar year. During the COVID pandemic, autonomous AI (IDx-DR, Digital Diagnostics, Coralville, IA) was deployed at multiple clinics, so that by 2021, some clinics had autonomous AI (AI sites), while others did not (non-AI sites). Overall adherence in 2019 (pre-AI) and 2021 (with AI) were compared, and stratified by demographics, using chi-square or Fisher’s exact test. Changes from 2019 to 2021 in odds of adherence within each subgroup were assessed for significant differences by site type (AI vs. non-AI), using logistic regression with a site type-by-time interaction term. From 2019 to 2021, the overall adherence rate increased from 42.6% to 55.5% at AI sites (1949 patients), and increased from 38.0% to 41.0% at non-AI sites (5379 patients). The increase in overall adherence rate at AI sites was significantly greater than that at non-AI sites (p<0.001). Various patient subgroups (M, F, ≥65 yo, <65 yo, Black, White, Medicare/military/commercial insurance) at AI sites demonstrated a significantly greater (p<0.05) compliance rate increase when compared to the same sub…
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The accuracy and low dilation requirement distinguish the AEYE-DS system from the other 2 commercially available systems.
“Though the enhanced efficiency and potential pitfalls in clinical workflows of taking 1 image versus 2 images per eye have yet to be studied, the accuracy and low dilation requirement distinguish the AEYE-DS system from the other 2 commercially available systems.”
From Autonomous Artificial Intelligence in Diabetic Retinopathy Testing—Lessons Learned on Successful Health System Adoption (2026) · cites this paper as [37]
AEYE-DS requires only 1 image per eye (macula-centered color fundus images with ≥45° field of view) for diagnosis and requires dilation in <1% of patients.
“The system requires only 1 image per eye (ie, macula-centered color fundus images with ≥45° field of view) for diagnosis and requires dilation in <1% of patients.37”
From Autonomous Artificial Intelligence in Diabetic Retinopathy Testing—Lessons Learned on Successful Health System Adoption (2026) · cites this paper as [37]
No citation lists for this paper
No citation lists for this paper yet — checked again daily.
Cited by 3
- Authors
- ALVIN LIU, T. Y., HUANG, J., LEHMANN, H., WOLF, R. M., CHANNA, R., & ABRÀMOFF, M. D.
- Affiliations
- Baltimore, MD, Madison, WI, Iowa City, IA
- Venue
- Diabetes
- Publisher
- American Diabetes Association
- Published
- 2023-06-20
- Volume
- 72
- Issue
- Supplement_1
- ISSN
- 0012-1797 (ISSN-L)
- OpenAlex
- W4381377927
- Cited by
- 3
- References
- 0
- Citations by year
- 2025: 1 · 2024: 1 · 2023: 1
- FWCI
- 0.50
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Diabetes (publishedVersion)
- Topics
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Retinal Imaging and Analysis
- Health Sciences › Medicine › Ophthalmology › Ophthalmology and Visual Health Research
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Keywords
- diabetic eye disease, diabetic retinopathy screening, AI autonomy, health equity, racial disparities, Black patients, primary health centers, COVID-19 pandemic, integrated health system, health inequalities reduction, telemedicine screening, community-based primary care, age and sex differences
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
