CoChat
Sign Up
Cited by 39Retinal Imaging and AnalysisArtificial Intelligence in Healthcare and EducationOphthalmology and Visual Health Research

Autonomous artificial intelligence for diabetic eye disease increases access and health equity in underserved populations

Open access
Huang, Jane J., Channa, Roomasa, Wolf, Risa M., Dong, Yiwen, Liang, Mavis, Wang, Jiangxia, Abramoff, Michael D., Liu, T. Y. Alvin·2024·npj Digital Medicine
DOI 10.1038/s41746-024-01197-3PMC PMC11263546

Published version of a preprint

Open this paper in CoChat — free accountOpen PDF

Abstract

Diabetic eye disease (DED) is a leading cause of blindness in the world. Annual DED testing is recommended for adults with diabetes, but adherence to this guideline has historically been low. In 2020, Johns Hopkins Medicine (JHM) began deploying autonomous AI for DED testing. In this study, we aimed to determine whether autonomous AI implementation was associated with increased adherence to annual DED testing, and how this differed across patient populations. JHM primary care sites were categorized as “non-AI” (no autonomous AI deployment) or “AI-switched” (autonomous AI deployment by 2021). We conducted a propensity score weighting analysis to compare change in adherence rates from 2019 to 2021 between non-AI and AI-switched sites. Our study included all adult patients with diabetes (>17,000) managed within JHM and has three major findings. First, AI-switched sites experienced a 7.6 percentage point greater increase in DED testing than non-AI sites from 2019 to 2021 ( p < 0.001). Second, the adherence rate for Black/African Americans increased by 12.2 percentage points within AI-switched sites but decreased by 0.6% points within non-AI sites ( p < 0.001), suggesting that autonomous AI deployment improved access to retinal evaluation for historically disadvantaged populations. Third, autonomous AI is associated with improved health equity, e.g. the adherence rate gap between Asian Americans and Black/African Americans shrank from 15.6% in 2019 to 3.5% in 2021. In summary, our results from real-world deployment in a large integrated healthcare system suggest that autonomous AI is associated with improvement in overall DED testing adherence, patient access, and health equity.

AI snapshot

Written from the paper’s full text.

TL;DR

In a real-world retrospective study at Johns Hopkins Medicine, deployment of autonomous AI for diabetic eye disease testing was associated with a 7.6 percentage point greater increase in annual DED testing adherence at AI-switched primary care sites than at non-AI sites from 2019 to 2021. The improvement was largest among historically disadvantaged groups, including Black/African American patients and Medicaid-insured patients, and narrowed some adherence gaps.

Key findings

  • AI-switched sites had a 7.6 percentage point greater increase in DED testing adherence than non-AI sites from 2019 to 2021 (95% CI 5.2–10.1, p < 0.001); in propensity score weighting, AI-switched sites increased by 7.4 percentage points (95% CI 5.2–9.5, p < 0.001) while non-AI sites decreased by 0.3 percentage points (95% CI −1.5–0.9, p = 0.653).
  • Overall adherence at AI-switched sites rose from 46.1% in 2019 to 54.5% in 2021 (+8.8 percentage points), while non-AI sites remained stable at 40.4% to 40.3% (+0.3 percentage points; p < 0.001).
  • Black/African American patients at AI-switched sites increased adherence by 12.2 percentage points, compared with a 0.6 percentage point decrease at non-AI sites (p < 0.001); Medicaid-insured patients increased by 13.7 percentage points and patients in the 2nd ADI quartile by 11.7 percentage points at AI-switched sites.
  • The adherence gap between Asian American and Black/African American patients at AI-switched sites shrank from 15.6% in 2019 to 3.5% in 2021.

Methods

  • Retrospective study approved by the Johns Hopkins School of Medicine IRB, including all patients with diabetes mellitus managed at Johns Hopkins Medicine primary care sites in 2019 and 2021; sites were categorized as “AI-switched” (no autonomous AI in 2019, deployed by 2021) or “non-AI.”
  • Primary outcome was adherence to annual DED testing, defined as receiving an autonomous AI exam or a dilated eye exam with an ophthalmologist.
  • Propensity score weighting with an inverse-probability-weighted regression adjustment model estimated average treatment effects across four site-year groups, adjusting for age, gender, race, ethnicity, preferred language, insurance, ADI quartile, family income, and geographic region; subgroup changes used Poisson regression with generalized estimating equations.

Limitations

  • Stated in the paper: the study is retrospective; nearly all patients lived in a metropolitan area, so findings may not generalize to micropolitan, small town, or rural populations; treatment heterogeneity was observed and its causes were not evaluated; a prospective randomized trial is needed to assess causality.
  • Evident from the text: the observational pre-post design may leave residual confounding despite propensity score weighting, and baseline differences existed between AI-switched and non-AI sites (e.g., income, ADI distribution, insurance, race), with the COVID pandemic also noted as a context for adherence trends.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41746-024-01197-3

Cites

23 of 23 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

13 of 27 cited sources checked · 6 supported · 0 flagged · accuracy not yet measured

  • Premisecitation

    Subsequent studies have further validated the accuracy, sensitivity, and specificity of this autonomous AI technology using ophthalmologists' reads as the reference standard.

    “Subsequent studies have further validated the accuracy, sensitivity, and specificity of this autonomous AI technology, using ophthalmologists’ reads as the reference standard9–11.”
    Quote in the paperNot a substantial claim3 of 3 sources checked
    • cites [9] Grzybowski A. et al. Diagnostic accuracy of Automated Diabetic Retinopathy Image Assessment Softwares: IDx-DR and MediosAI. Ophthalmic Res. https://doi.org/10.1159/000534098 (2023). source
    • cites [10] Shah A. et al. Validation of Automated Screening for Referable Diabetic Retinopathy With an Autonomous Diagnostic Artificial Intelligence System in a Spanish. Popul. J. Diabetes Sci. Technol. 15, 655–663 (2021). source
    • cites [11] van der Heijden, A. A. et al. Validation of automated screening for referable diabetic retinopathy with the IDx‐DR device in the Hoorn Diabetes Care System. Acta Ophthalmol. (Copenh.) 96, 63–68 (2018). source
  • Premisecitation

    Previous qualitative investigations identified misinformation about the importance of regular testing, logistic challenges with scheduling appointments, and anxiety as key barriers to annual eye exams.

    “Previous qualitative investigations have identified misinformation about the importance of regular testing, logistic challenges with scheduling appointments, and anxiety as key barriers to annual eye exams7.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [7] Kashim, R., Newton, P. & Ojo, O. Diabetic Retinopathy Screening: A Systematic Review on Patients’ Non-Attendance. Int. J. Environ. Res. Public. Health 15, 157 (2018). source

      The Results (§3.3; page not provided) identify misinformation about screening and anxiety as barriers, and report late screening invitations as a system-level reason for non-attendance. The review also describes annual screening in the Introduction, but it does not specifically identify difficulty scheduling appointments as a barrier. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
  • Premisecitation

    The autonomous AI system can autonomously analyze retinal images at the point of care with a sensitivity of 87.2% and specificity of 90.7% in a pivotal clinical trial against a prognostic standard.

    “This system can autonomously analyze images of the retina at the point of care, such as at primary care clinics, with a sensitivity of 87.2% and specificity of 90.7% in a pivotal clinical trial against a prognostic standard, i.e. patient outcome8.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [8] Abràmoff, M. D., Lavin, P. T., Birch, M., Shah, N. & Folk, J. C. Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices. Npj Digit. Med. 1, 39 (2018). source

      The Discussion reports 87.2% sensitivity and 90.7% specificity for the autonomous AI system in a primary-care setting. But the Abstract says the trial compared its results with graded retinal photographs and OCT images, not with patient outcomes; the claimed prognostic standard is unsupported. (LLM, conf=0.92)

      Experimental — accuracy not yet measured
  • Premisecitation

    In 2018, the FDA De Novo authorized the use of an autonomous artificial intelligence system (LumineticsCore, formerly IDx-DR) to diagnose diabetic eye disease.

    “In 2018, the FDA De Novo authorized the use of an autonomous artificial intelligence (AI) system (LumineticsCore® formerly known as IDx-DR, Digital Diagnostics, Coralville, IA) to diagnose DED.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [8] Abràmoff, M. D., Lavin, P. T., Birch, M., Shah, N. & Folk, J. C. Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices. Npj Digit. Med. 1, 39 (2018). source

      The Discussion states that FDA authorized the autonomous IDx-DR system to detect more-than-mild diabetic retinopathy in adults with diabetes who had not previously been diagnosed with diabetic retinopathy. Our copy does not identify the authorization as De Novo or say that IDx-DR was later renamed LumineticsCore; its stated diagnostic indication is also narrower than diabetic eye disease generally. (LLM, conf=0.87)

      Experimental — accuracy not yet measured
  • Premisecitation

    The African American population uses eye care services at much lower rates than White patients.

    “Prior studies have found that the African American population uses eye care services at much lower rates than White patients, even though Black patients with type 2 diabetes are significantly more likely to develop retinopathy than White patients13–16.”
    Quote in the paperSubstantial claim1 of 3 sources checked
    • cites [13] Sloan, F. A., Brown, D. S., Carlisle, E. S., Picone, G. A. & Lee, P. P. Monitoring Visual Status: Why Patients Do or Do Not Comply with Practice Guidelines. Health Serv. Res. 39, 1429–1448 (2004). source
    • cites [14] Harris, E. L., Sherman, S. H. & Georgopoulos, A. Black-white differences in risk of developing retinopathy among individuals with type 2 diabetes. Diabetes Care 22, 779–783 (1999). source
    • cites [15] Leske M. C. et al. Diabetic retinopathy in a black population. Ophthalmology 106, 1893–1899 (1999). source
  • Classified as probably not a substantial claim. 17
  • May combine two claims. 6

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Minority populations overall have consistently lower unadjusted eye examination rates than White populations.

    “In fact, minority populations overall have consistently lower unadjusted eye examination rates than White populations17.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [17] Shi, Q., Zhao, Y., Fonseca, V., Krousel-Wood, M. & Shi, L. Racial Disparity of Eye Examinations Among the U.S. Working-Age Population With Diabetes: 2002–2009. Diabetes Care 37, 1321–1328 (2014). source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The national ADI score is a 100-point scale that measures socioeconomic disadvantage based on data like a geographic region's housing quality, education, income, and employment.

    “The national ADI score is a 100-point scale that measures socioeconomic disadvantage based on data like a geographic region’s housing quality, education, income, and employment23.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [23] Hu, J., Kind, A. J. H. & Nerenz, D. Area Deprivation Index Predicts Readmission Risk at an Urban Teaching Hospital. Am. J. Med. Qual. 33, 493–501 (2018). source

      The Introduction describes ADI as a measure of neighborhood disadvantage, refined to the census-block-group level. But the Results report a national ADI range of 2.1–129.3, contrary to the claimed 100-point scale. The cited passages do not identify housing quality, education, income, and employment as its component data. Page numbers are unavailable in our copy. (LLM, conf=0.90)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A higher national ADI score correlates with increased overall socioeconomic disadvantage.

    “A higher national ADI score correlates with increased overall socioeconomic disadvantage.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [23] Hu, J., Kind, A. J. H. & Nerenz, D. Area Deprivation Index Predicts Readmission Risk at an Urban Teaching Hospital. Am. J. Med. Qual. 33, 493–501 (2018). source

  • Classified as probably not a substantial claim.

    Premisecitation

    Current guidelines recommend that patients with diabetes receive an annual eye examination.

    “current guidelines from the American Academy of Ophthalmology and American Diabetes Association recommend that patients with diabetes receive an annual eye examination2,3.”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites [2] Lin, K., Hsih, W., Lin, Y., Wen, C. & Chang, T. Update in the epidemiology, risk factors, screening, and treatment of diabetic retinopathy. J. Diabetes Investig. 12, 1322–1325 (2021). source

      The Screening section discusses eye-examination guidelines from the International Council of Ophthalmology and the American Diabetes Association, and describes routine annual screening for type 1 diabetes as a comparator. It does not say that those guidelines recommend annual examinations for all patients with diabetes, or identify the American Academy of Ophthalmology as a guideline source. Page numbers are unavailable in our copy. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [3] Kuo J. et al. Factors associated with adherence to screening guidelines for diabetic retinopathy among low-income metropolitan patients. Mo Med. 117, 258–264 (2020).

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Black patients with type 2 diabetes are significantly more likely to develop retinopathy than White patients.

    “Prior studies have found that the African American population uses eye care services at much lower rates than White patients, even though Black patients with type 2 diabetes are significantly more likely to develop retinopathy than White patients13–16.”
    Quote in the paperNot a substantial claim1 of 3 sources checked
    • cites [13] Sloan, F. A., Brown, D. S., Carlisle, E. S., Picone, G. A. & Lee, P. P. Monitoring Visual Status: Why Patients Do or Do Not Comply with Practice Guidelines. Health Serv. Res. 39, 1429–1448 (2004). source

      The Discussion and Conclusions (page unavailable) report that African Americans used eye care services at much lower rates than whites, supporting the first part of the citing sentence. Neither that section nor the other retrieved passages report that Black patients with type 2 diabetes are significantly more likely to develop retinopathy than White patients; the paper examines eye-exam use, not that comparative risk. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [14] Harris, E. L., Sherman, S. H. & Georgopoulos, A. Black-white differences in risk of developing retinopathy among individuals with type 2 diabetes. Diabetes Care 22, 779–783 (1999). source
    • cites [15] Leske M. C. et al. Diabetic retinopathy in a black population. Ophthalmology 106, 1893–1899 (1999). source

  • Classified as probably not a substantial claim.

    Premisecitation

    The adherence rate is even lower (30-40%) in smaller health systems and low-income metropolitan patient populations.

    “This adherence rate is even lower (30–40%) in smaller health systems and low-income metropolitan patient populations3,6.”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites [3] Kuo J. et al. Factors associated with adherence to screening guidelines for diabetic retinopathy among low-income metropolitan patients. Mo Med. 117, 258–264 (2020).
    • cites [6] Lock L. J. et al. Analysis of Health System Size and Variability in Diabetic Eye Disease Screening in Wisconsin. JAMA Netw. Open 5, e2143937 (2022). source

      The Results (page unavailable) report less screening in smaller health systems and an overall range of 31.8%–73.0%, but do not establish a 30%–40% rate specifically for smaller systems. The paper does not report screening rates for low-income metropolitan patients; its Discussion (page unavailable) instead identifies rural primary care clinics as having lower screening. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data

    We observed treatment heterogeneity with improvement in adherence rate even after autonomous AI deployment.

    “Fourth, we observed treatment heterogeneity with improvement in adherence rate even after autonomous AI deployment.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Diabetic eye disease is a leading cause of blindness in the world.

    “Diabetic eye disease (DED) is a leading cause of blindness in the world.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [1] Cheung, N., Mitchell, P. & Wong, T. Y. Diabetic retinopathy. Lancet 376, 124–136 (2010). source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Diabetic eye disease affects a third of people with diabetes mellitus and is a leading cause of blindness and visual impairment in working-aged adults in the developed world.

    “Diabetic eye disease (DED) affects a third of people with diabetes mellitus (DM) and is a leading cause of blindness and visual impairment in working-aged adults in the developed world1.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [1] Cheung, N., Mitchell, P. & Wong, T. Y. Diabetic retinopathy. Lancet 376, 124–136 (2010). source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The nationwide average adherence rate was 58.3%.

    “which was lower than the nationwide average of 58.3%12”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [12] U.S. Department of Health and Human Services. Healthy People 2030. https://health.gov/healthypeople/objectives-and-data/browse-objectives/diabetes/increase-proportion-adults-diabetes-who-have-yearly-eye-exam-d-04/data (2020).

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Classified as probably not a substantial claim.

    Premisecitation

    Annual screenings allow for early diagnosis and treatment that can help prevent severe vision loss.

    “These annual screenings allow for early diagnosis and treatment that can help prevent severe vision loss4.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [4] Bragge P. Screening for Presence or Absence of Diabetic Retinopathy: A Meta-analysis. Arch. Ophthalmol. 129, 435–444 (2011). source

  • May combine two claims.

    Interpretationown data

    Among AI-switched sites, the patient populations that experienced substantial improvement in adherence rate included Black or African American patients, patients with Medicaid insurance coverage, and patients with high ADI scores.

    “Among the AI-switched sites, the patient populations that experienced substantial improvement in adherence rate included Black or African American patients, patients with Medicaid insurance coverage, and patients with high ADI scores.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    The study was retrospective.

    “This was a retrospective study approved by the Institutional Review Board of the Johns Hopkins School of Medicine.”
    • Also stated in: § Discussion · plain
      “Our study is limited by its retrospective nature”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    A previous study in the United States found that only 50-60% of Medicare beneficiaries with diabetes mellitus received annual eye exams.

    “A previous study in the United States found that only 50–60% of Medicare beneficiaries with DM received annual eye exams5.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [5] Lee, P. P., Feldman, Z. W., Ostermann, J., Brown, D. S. & Sloan, F. A. Longitudinal rates of annual eye examinations of persons with diabetes and chronic eye diseases. Ophthalmology 110, 1952–1959 (2003). source

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Robust standard errors estimate clustering on participants was used to account for patients who were seen in both 2019 and 2021.

    “Robust standard errors estimate clustering on participants was used to account for patients who were seen in both 2019 and 2021.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Though autonomous AI improved access to retinal evaluation for the most disadvantaged patient groups and reduced care gaps, such improvement was not universal.

    “Though autonomous AI improved access to retinal evaluation for the most disadvantaged patient groups and reduced care gaps, such improvement was not universal.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Covariates balance was checked after inverse probability weighting to ensure appropriate balance on all covariates among the four groups.

    “The covariates balance was checked after the inverse probability weighting to ensure appropriate balance on all covariates among the four groups.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The propensity score weighting and measured covariates account for all differences between AI-switched and non-AI sites that affect adherence, i.e. there is no unmeasured confounding.

  • May combine two claims.

    Interpretationown data

    Nearly all patients included in the study live in a metropolitan area, so the observations may not generalize to patient populations living in micropolitan, small town, and rural residences.

    “Our study is limited by its retrospective nature and the fact that nearly all patients included in our study live in a metropolitan area, so our observations may not generalize to patient populations living in micropolitan, small town, and rural residences.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A previous focus group found that for many African American patients, transportation, lack of free time, and inconvenience of eye exam served as main barriers for annual diabetic eye disease screening.

    “A previous focus group found that for many African American patients, transportation, lack of free time, and inconvenience of eye exam served as main barriers for annual DED screening16.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [16] Fathy, C., Patel, S., Sternberg, P. & Kohanim, S. Disparities in Adherence to Screening Guidelines for Diabetic Retinopathy in the United States: a Comprehensive Review and Guide for Future Directions. Semin. Ophthalmol. 31, 364–377 (2016). source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductnone given

    This was a retrospective study approved by the Institutional Review Board of the Johns Hopkins School of Medicine.

    “This was a retrospective study approved by the Institutional Review Board of the Johns Hopkins School of Medicine.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    All research adhered to the tenets of the Declaration of Helsinki.

    “All research adhered to the tenets of the Declaration of Helsinki.”
    Quote in the paperNot a substantial claim
21 of 21

  • The paper relies on this without support.

    The propensity score weighting and measured covariates account for all differences between AI-switched and non-AI sites that affect adherence, i.e. there is no unmeasured confounding.

    “we used inverse probability to assign weights to patients based on social determinants of health to ensure comparability across the different types of sites and years.”
    “We employed propensity score weighting methods to address the inherent limitations of an observational study and are reassured that our analyses showed a significant impact of autonomous AI on DED testing adherence rates.”

  • The paper relies on this without support.

    In the absence of autonomous AI deployment, adherence trends at AI-switched and non-AI sites would have been parallel.

    “The difference-in-difference was also estimated from this model.”
    “From 2019 to 2021, i.e. coming out of the COVID pandemic, which caused widespread adherence issues, we observed a substantial increase in adherence rate among AI-switched sites, while the adherence rate among non-AI sites remained unchanged.”

  • The paper relies on this without support.

    The COVID pandemic caused widespread adherence issues.

    “From 2019 to 2021, i.e. coming out of the COVID pandemic, which caused widespread adherence issues, we observed a substantial increase in adherence rate among AI-switched sites, while the adherence rate among non-AI sites remained unchanged.”
3 of 3

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?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • 2019 AI-switched American Indian or Alaska Native percentage is misrounded

    Minor

    The reported 25 patients out of 5,505 are 0.454%, which rounds to 0.5% at the table's precision, not 0.4%.

    Reported 0.4 · recomputed 0.454133

  • 2021 insurance percentages do not match Table 1

    Minor

    Table 1 gives 9,385 commercial/other patients and 4,966 pure Medicare patients out of 17,590 in 2021. These are 53.4% and 28.2%, respectively, rather than 53.1% and 28.1%.

  • 2021 non-AI self-pay percentage is misrounded

    Minor

    The reported 105 self-pay patients out of 12,010 are 0.874%, which rounds to 0.9%, not 0.8%.

    Reported 0.8 · recomputed 0.874271

  • 2021 gender and age percentages do not match Table 1

    Minor

    Table 1 reports 9,029 female patients and 12,562 patients under 65 out of 17,590 in 2021. Those are 51.3% and 71.4%, respectively, rather than the stated 51.1% and 71.1%.

  • 2021 racial percentages do not match Table 1

    Minor

    The 2021 counts in Table 1 give 8,366 White patients and 6,555 Black or African American patients out of 17,590. These correspond to 47.6% and 37.3%, not 47.9% and 37.1%.

  • 2019 AI-switched self-pay percentage is misrounded

    Minor

    The reported 64 self-pay patients out of 5,505 are 1.163%, which rounds to 1.2%, not 1.1%.

    Reported 1.1 · recomputed 1.162579

6 of 6

Cites

Flagged first, then by citations

1 of the 24 references couldn’t be matched to a known paper.

Cited by

Sorted by citations

Related

Authors
Huang, J. J., Channa, R., Wolf, R. M., Dong, Y., Liang, M., Wang, J., Abramoff, M. D., & Liu, T. Y. A.
Affiliations
  • Johns Hopkins University (US)
  • University of Wisconsin–Madison (US)
Corresponding author
Liu, T. Y. Alvin
ORCID
Venue
npj Digital Medicine
Publisher
Springer Science and Business Media LLC
Published
2024-07-22
Volume
7
Issue
1
Pages
196-196
ISSN
2398-6352 (ISSN-L)
DOI
10.1038/s41746-024-01197-3
PMC
PMC11263546
OpenAlex
W4400876649
Semantic Scholar
63e389079c0aad489095ed2e90a7c5746b45e0ee
Cited by
39
References
24
Citations by year
2026: 17 · 2025: 16 · 2024: 6
FWCI
8.97
Type
journal article
Language
en
Update notices
correction (2024-08-23)
Open access
Licence
Locations
  • npj Digital Medicine (publishedVersion, cc-by)
  • Research Square (acceptedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
Funders
Research to Prevent Blindness Career Advancement Award
Topics
  • Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Retinal Imaging and Analysis
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Health Sciences › Medicine › Ophthalmology › Ophthalmology and Visual Health Research
Keywords
diabetic eye disease, AI autonomy, retinal screening, health equity, underserved populations, Black patients, Asian American patients, diabetes mellitus, primary care, integrated health system

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.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-15.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-15.

Cited sources are checked by gpt-6-sol.

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