Autonomous Artificial Intelligence Increases Access and Health Equity in Underserved Populations with Diabetes.

Diabetic eye disease (DED) is a leading cause of blindness in the world. Early detection and treatment of DED have been shown to be both sight-saving and cost-effective. As such, annual testing for DED is recommended for adults with diabetes and is a Healthcare Effectiveness Data and Information Set (HEDIS) measure. However, adherence to this guideline has historically been low, and access to this sight-saving intervention has particularly been limited for specific populations, such as Black or African American patients. In 2018, the US Food and Drug Agency (FDA) De Novo cleared autonomous artificial intelligence (AI) for diagnosing DED in a primary care setting. In 2020, Johns Hopkins Medicine (JHM), an integrated healthcare system with over 30 primary care sites, began deploying autonomous AI for DED testing in some of its primary care clinics. In this retrospective study, we aimed to determine whether autonomous AI implementation was associated with increased adherence to annual DED testing, and whether this was different for specific populations. JHM primary care sites were categorized as "non-AI" sites (sites with no autonomous AI deployment over the study period and where patients are referred to eyecare for DED testing) or "AI-switched" sites (sites that did not have autonomous AI testing in 2019 but did by 2021). We conducted a difference-in-difference analysis using a logistic regression model 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 managed within our health system (17,674 patients for the 2019 cohort and 17,590 patients for the 2021 cohort) and has three major findings. First, after controlling for a wide range of potential confounders, our regression analysis demonstrated that the odds ratio of adherence at AI-switched sites was 36% higher than that of non-AI sites, suggesting that there was a higher increase in DED testing between 2019 and 2021 at AI-switched sites than at non-AI sites. Second, our data suggested autonomous AI improved access for historically disadvantaged populations. The adherence rate for Black/African Americans increased by 11.9% within AI-switched sites whereas it decreased by 1.2% within non-AI sites over the same time frame. Third, the data suggest that autonomous AI improved health equity by closing care gaps. For example, in 2019, a large adherence rate gap existed between Asian Americans and Black/African Americans (61.1% vs. 45.5%). This 15.6% gap shrank to 3.5% by 2021. In summary, our real-world deployment results in a large integrated healthcare system suggest that autonomous AI improves adherence to a HEDIS measure, patient access, and health equity for patients with diabetes - particularly in historically disadvantaged patient groups. While our findings are encouraging, they will need to be replicated and validated in a prospective manner across more diverse settings.

Use of Diabetes Technologies and Retinopathy in Adults With Type 1 Diabetes.

Importance: Diabetic retinopathy (DR) is a complication of diabetes that can lead to vision loss. Outcomes of continuous glucose monitoring (CGM) and insulin pump use in DR are not well understood. Objective: To assess the use of CGM, insulin pump, or both, and DR and proliferative diabetic retinopathy (PDR) in adults with type 1 diabetes (T1D). Design, Setting, and Participants: A retrospective cohort study of adults with T1D in a tertiary diabetes center and ophthalmology center was conducted from 2013 to 2021, with data analysis performed from June 2022 to April 2023. Exposure: Use of diabetes technologies, including insulin pump, CGM, and both CGM and insulin pump. Main Outcomes and Measures: The primary outcome was development of DR or PDR. A secondary outcome was the progression of DR for patients in the longitudinal cohort. Multivariable logistic regression models assessed for development of DR and PDR and association with CGM and insulin pump use. Results: A total of 550 adults with T1D were included (median age, 40 [IQR, 28-54] years; 54.4% female; 24.5% Black or African American; and 68.4% White), with a median duration of diabetes of 20 (IQR, 10-30) years, and median hemoglobin A1c (HbA1c) of 7.8% (IQR, 7.0%-8.9%). Overall, 62.7% patients used CGM, 58.2% used an insulin pump, and 47.5% used both; 44% (244 of 550) of the participants had DR at any point during the study. On univariate analysis, CGM use was associated with lower odds of DR and PDR, and CGM with pump was associated with lower odds of PDR (all P < .05), compared with no CGM use. Multivariable logistic regression adjusting for age, sex, race and ethnicity, diabetes duration, microvascular and macrovascular complications, insurance type, and mean HbA1c, showed that CGM was associated with lower odds of DR (odds ratio [OR], 0.52; 95% CI, 0.32-0.84; P = .008) and PDR (OR, 0.42; 95% CI, 0.23-0.75; P = .004), compared with no CGM use. In the longitudinal analysis of participants without baseline PDR, 79 of 363 patients (21.8%) had progression of DR during the study. Conclusions and Relevance: In this cohort study of adults with T1D, CGM use was associated with lower odds of developing DR and PDR, even after adjusting for HbA1c. These findings suggest that CGM may be useful for diabetes management to mitigate risk for DR and PDR.

Comparison of Standard 7-Field, Clarus, and Optos Ultrawidefield Imaging Systems for Diabetic Retinopathy (COCO Study).

Purpose: The purpose of this study was to compare diabetic retinopathy (DR) severity levels assessed from 7 standard-field stereoscopic color photographs on a 35° fundus camera to both Clarus and Optos ultrawidefield color images. Design: Cross-sectional, comparative imaging study. Participants: Participants with DR imaged at a single-center retina practice. Methods: Participants were imaged on 3 cameras at a single visit with the Topcon 35° fundus camera, Clarus, and Optos. The DR Severity Scale (DRSS) level was determined within the 7-field (7F) area of each image set using the ETDRS scale. An additional global DRSS was assigned for both Clarus and Optos images using the entire visible retina. Weighted kappa (wκ) measured the agreement between cameras. Main Outcome Measures: The primary outcome was a 3-way comparison of DRSS level within the 7F area imaged on the 3 cameras. Secondary outcomes included a comparison of the DRSS obtained with standard 7F imaging to the global DRSS of Clarus and Optos and a comparison of the global DRSS between Clarus and Optos only. Results: Ninety-seven eyes (50 participants) were evaluated. Agreement within 1-step of ETDRS levels between standard 7F imaging and Clarus 7F was 90.1% (wκ = 0.65), and with Optos 7F in 85.9%, (wκ = 0.58). Agreement within 1-step between standard 7F imaging and Clarus global was 88.9% of eyes (wκ = 0.63), and Optos global was 85.7%, (wκ = 0.54). Agreement between Clarus and Optos global DR level within 1-step was 89.1% (wκ = 0.68). Intergrader agreement for the 7F ETDRS level was 96% for standard 7F imaging, 98% for Clarus, and 95.5% for Optos. Conclusions: These findings suggest that when evaluating the 7F area on Clarus and Optos, DR severity grades are comparable to standard 7F imaging. However, it is important to understand the unique attributes and differences of each fundus camera when changing the type of system used in a clinical setting due to upgrading equipment. Additionally, if the facility has access to > 1 device, there should not be an exchange between cameras for the same patient. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Autonomous artificial intelligence increases screening and follow-up for diabetic retinopathy in youth: the ACCESS randomized control trial.

Diabetic retinopathy can be prevented with screening and early detection. We hypothesized that autonomous artificial intelligence (AI) diabetic eye exams at the point-of-care would increase diabetic eye exam completion rates in a racially and ethnically diverse youth population. AI for Children's diabetiC Eye ExamS (NCT05131451) is a parallel randomized controlled trial that randomized youth (ages 8-21 years) with type 1 and type 2 diabetes to intervention (autonomous artificial intelligence diabetic eye exam at the point of care), or control (scripted eye care provider referral and education) in an academic pediatric diabetes center. The primary outcome was diabetic eye exam completion rate within 6 months. The secondary outcome was the proportion of participants who completed follow-through with an eye care provider if deemed appropriate. Diabetic eye exam completion rate was significantly higher (100%, 95%CI: 95.5%, 100%) in the intervention group (n = 81) than the control group (n = 83) (22%, 95%CI: 14.2%, 32.4%)(p < 0.001). In the intervention arm, 25/81 participants had an abnormal result, of whom 64% (16/25) completed follow-through with an eye care provider, compared to 22% in the control arm (p < 0.001). Autonomous AI increases diabetic eye exam completion rates in youth with diabetes.

A New Approach to Staging Diabetic Eye Disease: Staging of Diabetic Retinal Neurodegeneration and Diabetic Macular Edema.

Topic: The goal of this review was to summarize the current level of evidence on biomarkers to quantify diabetic retinal neurodegeneration (DRN) and diabetic macular edema (DME). Clinical relevance: With advances in retinal diagnostics, we have more data on patients with diabetes than ever before. However, the staging system for diabetic retinal disease is still based only on color fundus photographs and we do not have clear guidelines on how to incorporate data from the relatively newer modalities into clinical practice. Methods: In this review, we use a Delphi process with experts to identify the most promising modalities to identify DRN and DME. These included microperimetry, full-field flash electroretinogram, spectral-domain OCT, adaptive optics, and OCT angiography. We then used a previously published method of determining the evidence level to complete detailed evidence grids for each modality. Results: Our results showed that among the modalities evaluated, the level of evidence to quantify DRN and DME was highest for OCT (level 1) and lowest for adaptive optics (level 4). Conclusion: For most of the modalities evaluated, prospective studies are needed to elucidate their role in the management and outcomes of diabetic retinal diseases. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Risk Factors for Nondiagnostic Imaging in a Real-World Deployment of Artificial Intelligence Diabetic Retinal Examinations in an Integrated Healthcare System: Maximizing Workflow Efficiency Through Predictive Dilation.

OBJECTIVE: In the pivotal clinical trial that led to Food and Drug Administration De Novo "approval" of the first fully autonomous artificial intelligence (AI) diabetic retinal disease diagnostic system, a reflexive dilation protocol was used. Using real-world deployment data before implementation of reflexive dilation, we identified factors associated with nondiagnostic results. These factors allow a novel predictive dilation workflow, where patients most likely to benefit from pharmacologic dilation are dilated a priori to maximize efficiency and patient satisfaction. METHODS: Retrospective review of patients who were assessed with autonomous AI at Johns Hopkins Medicine (8/2020 to 5/2021). We constructed a multivariable logistic regression model for nondiagnostic results to compare characteristics of patients with and without diagnostic results, using adjusted odds ratio (aOR). P < .05 was considered statistically significant. RESULTS: Of 241 patients (59% female; median age = 59), 123 (51%) had nondiagnostic results. In multivariable analysis, type 1 diabetes (T1D, aOR = 5.82, 95% confidence interval [CI]: 1.45-23.40, P = .01), smoking (aOR = 2.86, 95% CI: 1.36-5.99, P = .005), and age (every 10-year increase, aOR = 2.12, 95% CI: 1.62-2.77, P < .001) were associated with nondiagnostic results. Following feature elimination, a predictive model was created using T1D, smoking, age, race, sex, and hypertension as inputs. The model showed an area under the receiver-operator characteristics curve of 0.76 in five-fold cross-validation. CONCLUSIONS: We used factors associated with nondiagnostic results to design a novel, predictive dilation workflow, where patients most likely to benefit from pharmacologic dilation are dilated a priori. This new workflow has the potential to be more efficient than reflexive dilation, thus maximizing the number of at-risk patients receiving their diabetic retinal examinations.

Factors Influencing Eye Screening Adherence Among Latinx Patients With Diabetes: A Qualitative Study.

Purpose: Latinx populations have the highest rates of visual impairment and blindness of any ethnic group in the United States, with most cases of diabetic retinopathy remaining undiagnosed. We aimed to identify factors influencing adherence with diabetic eye screening in Latinx communities. Methods: We conducted semistructured individual interviews with adult Latinx patients in Dane County, WI. Interviews were transcribed verbatim, translated from Spanish to English, and analyzed using QSR NVivo software. We performed both inductive open coding and deductive coding using the National Institute on Minority Health and Health Disparities Research Framework, as well as the Campbell and Egede Model. Results: All participants (n = 20) self-identified as Latinx and were diagnosed with type 2 diabetes. The mean age was 61.5 years (range 33-79 years). Most participants were uninsured (60%), self-reported low or moderate health literacy (60%), and preferred to speak Spanish during their clinic appointments (75%). Individual-level barriers to diabetic eye screening included limited eye health literacy, lack of insurance coverage, and low self-efficacy with diabetes management. Health system-level facilitators included a recommendation to obtain eye screening from a primary care provider and the use of nonwritten forms of patient education. Community-level barriers included social isolation, concerns about inconveniencing others, machismo, and immigration status. Conclusions: We identified several health system- and community-level factors, in addition to individual-level factors, influencing adherence with diabetic eye screening in Latinx communities. Translational Relevance: Strategies addressing these factors may enhance the effectiveness of interventions to prevent blindness from diabetes and contribute to advancing health equity in Latinx communities.

Perspectives of Latinx Patients with Diabetes on Teleophthalmology, Artificial Intelligence-Based Image Interpretation, and Virtual Care: A Qualitative Study.

Background: Latinx populations in the United States bear a disproportionate burden of diabetic eye disease. Teleophthalmology with and without artificial intelligence (AI)-based image interpretation are validated methods for diabetic eye screening, but limited literature exists on patient perspectives. This study aimed at understanding the perspectives of Latinx patients with diabetes on teleophthalmology, AI-based image interpretation, and general virtual care to prevent avoidable blindness in this population. Methods: We conducted semi-structured, individual interviews with 20 Latinx patients with diabetes at an urban, federally qualified health center in Madison, WI. Interviews were transcribed verbatim, professionally translated from Spanish to English, and analyzed using both inductive open coding and deductive coding. Results: Most participants had no prior experience with teleophthalmology but did have experience with virtual care. Participants expressed a preference for teleophthalmology compared with traditional in-person dilated eye exams but were willing to obtain whichever method of screening was recommended by their primary care clinician. They also strongly preferred having human physician oversight in image review compared with having images interpreted solely using AI. Many participants preferred in-person clinic visits to virtual health care due to the ability to have a more thorough physical exam, as well as for improved non-verbal communication with their clinician. Discussion: Leveraging primary care providers' recommendations, human oversight of AI-based image interpretation, and improving communication may enhance acceptance and utilization of teleophthalmology, AI, and virtual care by Latinx patients. Conclusions: Understanding Latinx patient perspectives may contribute toward the development of more effective telemedicine interventions to enhance health equity in Latinx communities.

Synthetic OCT-A blood vessel maps using fundus images and generative adversarial networks.

Vessel segmentation in fundus images permits understanding retinal diseases and computing image-based biomarkers. However, manual vessel segmentation is a time-consuming process. Optical coherence tomography angiography (OCT-A) allows direct, non-invasive estimation of retinal vessels. Unfortunately, compared to fundus images, OCT-A cameras are more expensive, less portable, and have a reduced field of view. We present an automated strategy relying on generative adversarial networks to create vascular maps from fundus images without training using manual vessel segmentation maps. Further post-processing used for standard en face OCT-A allows obtaining a vessel segmentation map. We compare our approach to state-of-the-art vessel segmentation algorithms trained on manual vessel segmentation maps and vessel segmentations derived from OCT-A. We evaluate them from an automatic vascular segmentation perspective and as vessel density estimators, i.e., the most common imaging biomarker for OCT-A used in studies. Using OCT-A as a training target over manual vessel delineations yields improved vascular maps for the optic disc area and compares to the best-performing vessel segmentation algorithm in the macular region. This technique could reduce the cost and effort incurred when training vessel segmentation algorithms. To incentivize research in this field, we will make the dataset publicly available to the scientific community.

Systemic Adverse Events Among Patients With Diabetes Treated With Intravitreal Anti-Vascular Endothelial Growth Factor Injections.

Importance: Anti-vascular endothelial growth factor (VEGF) agents are currently the mainstay of treatment for diabetic retinopathy (DR). Although effective, data on their systemic safety remains inconclusive, particularly in high-risk patient groups. Objective: To explore the systemic safety of intravitreal anti-VEGF agents among patients with diabetes. Design, Setting, and Participants: This was a retrospective, longitudinal population-based analysis of the Corporate Data Warehouse, a large-scale database of patients within the US Veteran Health Affairs. All patients 18 years and older with type 2 diabetes who were seen at any Veterans Affairs health care facility in the US between January 1, 2011, and December 31, 2012, were identified. Data were then extracted on incident systemic adverse events among this patient cohort from January 1, 2013, to December 31, 2017. All individuals with diabetes who did and did not receive anti-VEGF injections were included. Patients with a history of prior systemic adverse events and those who received an intravitreal injection between January 1, 2011, and December 31, 2012, were excluded. Data were analyzed from October 2019 to March 2023. Exposure: Anti-VEGF injection. Main Outcomes and Measures: Proportion of patients with any incident systemic adverse event, acute myocardial infarction, cardiovascular disease, or kidney disease at 1-, 3-, and 5-year follow-up. Results: A total of 1 731 782 patients (mean [SD] age, 63.8 [12.3] years; 1 656 589 [95.7%] male) with type 2 diabetes were included. DR was present in 476 013 (27.5%), and 14 022 (0.8%) received anti-VEGF injections. Of the total number of patients with type 2 diabetes, 321 940 (18.6%) developed systemic adverse events between 2013 and 2017. The 5-year cumulative incidence of any systemic adverse event was 37.0% (5187/14 022) in the injection group vs 18.4% (316 753/1 717 760) in the noninjection group (P < .001). Anti-VEGF injections were independently associated with a higher likelihood of developing any systemic adverse event (odds ratio, 1.8; 95% CI, 1.7-1.9) when controlling for age, race, sex, ethnicity, tobacco use, severity of DR, Deyo-Charlson Comorbidity Index score, mean hemoglobin A1c, total number of injections, and statin use. Conclusion and Relevance: In this study, intravitreal anti-VEGF injections were independently associated with a higher likelihood of systemic adverse events among patients with diabetes.

Effectiveness of artificial intelligence screening in preventing vision loss from diabetes: a policy model.

The effectiveness of using artificial intelligence (AI) systems to perform diabetic retinal exams ('screening') on preventing vision loss is not known. We designed the Care Process for Preventing Vision Loss from Diabetes (CAREVL), as a Markov model to compare the effectiveness of point-of-care autonomous AI-based screening with in-office clinical exam by an eye care provider (ECP), on preventing vision loss among patients with diabetes. The estimated incidence of vision loss at 5 years was 1535 per 100,000 in the AI-screened group compared to 1625 per 100,000 in the ECP group, leading to a modelled risk difference of 90 per 100,000. The base-case CAREVL model estimated that an autonomous AI-based screening strategy would result in 27,000 fewer Americans with vision loss at 5 years compared with ECP. Vision loss at 5 years remained lower in the AI-screened group compared to the ECP group, in a wide range of parameters including optimistic estimates biased toward ECP. Real-world modifiable factors associated with processes of care could further increase its effectiveness. Of these factors, increased adherence with treatment was estimated to have the greatest impact.

Associations Between HbA1c Across the Normal Range, Diagnosed, and Undiagnosed Diabetes and Retinal Layer Thickness in UK Biobank Cohort.

Purpose: The purpose of this study was to investigate the association between glycated hemoglobin (HbA1c) levels and retinal sub-layer thicknesses in people with and without diabetes. Methods: We included 41,453 UK Biobank participants aged 40 to 69 years old. Diabetes status was defined by self-report of diagnosis or use of insulin. Participants were categorized into groups: (1) those with HbA1c <48 mmol/mol were subdivided into quintiles according to normal range of HbA1c; (2) those previously diagnosed with diabetes with no evidence of diabetic retinopathy; and (3) undiagnosed diabetes: >48 mmol/mol. Total macular and retinal sub-layer thicknesses were derived from spectral-domain optical coherence tomography (SD-OCT) images. Multivariable linear regression was used to evaluate the associations between diabetes status and retinal layer thickness. Results: Compared with participants in the second quintile of the normal HbA1c range, those in the fifth quintile had a thinner photoreceptor layer thickness (-0.33 µm, P = 0.006). Participants with diagnosed diabetes had a thinner macular retinal nerve fiber layer (mRNFL; -0.58 µm, P < 0.001), photoreceptor layer thickness (-0.94 µm, P < 0.001), and total macular thickness (-1.61 µm, P < 0.001), whereas undiagnosed diabetes participants had a reduced photoreceptor layer thickness (-1.22 µm, P = 0.009) and total macular thickness (-2.26 µm, P = 0.005). Compared to participants without diabetes, those with diabetes had a thinner mRNFL (-0.50 µm, P < 0.001), photoreceptor layer thickness (-0.77 µm, P < 0.001), and total macular thickness (-1.36 µm, P < 0.001). Conclusions: Participants with higher HbA1c in the normal range had marginally thinner photoreceptor thickness, whereas those with diabetes (including undiagnosed diabetes) had meaningfully thinner retinal sublayer and total macular thickness. Translational Relevance: We showed that early retinal neurodegeneration occurs in people whose HbA1c levels are below the current diabetes diagnostic threshold; this might impact the management of pre-diabetes individuals.

Retractable Urology Three-Pronged Grasping Forceps for Removal of Large, Non-Magnetic Intraocular Foreign Bodies.

PURPOSE: To demonstrate the use of urology retractable three-pronged grasping forceps in the removal of a large, round, and non-magnetic intraocular foreign body (IOFB) that was difficult to remove with other surgical instruments. METHODS: Extraction of a 3.0 mm lead shot pellet embedded in vitreous hemorrhage was attempted with multiple surgical instruments including an intraocular magnet, IOFB forceps, and two tools designed for urology stone removal: a three-pronged grasping forceps and a nitinol basket extractor. RESULTS: Due to the round and smooth surface, large size, and non-magnetic nature of the IOFB, extraction was challenging and failed with multiple other surgical instruments. The wide and secure grasp of the grasping forceps allowed for swift IOFB extraction without iatrogenic injury to the retina. CONCLUSION: The grasping forceps offer an effective and safe method for removal of large, round, and non-magnetic IOFBs.

Evaluation of retinal nerve fibre layer thickness as a possible measure of diabetic retinal neurodegeneration in the EPIC-Norfolk Eye Study.

BACKGROUND/AIMS: Markers to clinically evaluate structural changes from diabetic retinal neurodegeneration (DRN) have not yet been established. To study the potential role of peripapillary retinal nerve fibre layer (pRNFL) thickness as a marker for DRN, we evaluated the relationship between diabetes, as well as glycaemic control irrespective of diabetes status and pRNFL thickness. METHODS: Leveraging data from a population-based cohort, we used general linear mixed models (GLMMs) with a random intercept for patient and eye to assess the association between pRNFL thickness (measured using GDx) and demographic, systemic and ocular parameters after adjusting for typical scan score. GLMMs were also used to determine: (1) the relationship between: (A) glycated haemoglobin (HbA1c) irrespective of diabetes diagnosis and pRNFL thickness, (B) diabetes and pRNFL thickness and (2) which quadrants of pRNFL may be affected in participants with diabetes and in relation to HbA1c. RESULTS: 7076 participants were included. After controlling for covariates, inferior pRNFL thickness was 0.94 µm lower (95% CI -1.28 µm to -0.60 µm), superior pRNFL thickness was 0.83 µm lower (95% CI -1.17 µm to -0.49 µm) and temporal pRNFL thickness was 1.33 µm higher (95% CI 0.99 µm to 1.67 µm) per unit increase in HbA1c. Nasal pRNFL thickness was not significantly associated with HbA1c (p=0.23). Similar trends were noted when diabetes was used as the predictor. CONCLUSION: Superior and inferior pRNFL was significantly thinner among those with higher HbA1c levels and/or diabetes, representing areas of the pRNFL that may be most affected by diabetes.

Factors Associated with Follow-Up Adherence After Teleophthalmology for Diabetic Eye Screening Before and During the COVID-19 Pandemic.

Abstract Background: Follow-up adherence with in-person care is critical for achieving improved clinical outcomes in telemedicine screening programs. We sought to quantify the impact of the COVID-19 pandemic upon follow-up adherence and factors associated with follow-up adherence after teleophthalmology for diabetic eye screening. Methods: We retrospectively reviewed medical records of adults screened in a clinical teleophthalmology program at urban and rural primary care clinics between May 2015 and December 2020. We defined follow-up adherence as medical record documentation of an in-person eye exam within 1 year among patients referred for further care. Regression models were used to identify factors associated with follow-up adherence. Results: Among 948 patients, 925 (97.6%) had health insurance and 170 (17.9%) were referred for follow-up. Follow-up adherence declined from 62.7% (n = 52) prepandemic to 46.0% (n = 40) during the pandemic (p = 0.04). There was a significant decline in follow-up adherence among patients from rural (p < 0.001), but not urban (p = 0.72) primary care clinics. Higher median household income (odds ratio [OR] 1.68, 95% confidence interval [CI]: 1.19-2.36) and obtaining care from an urban clinic (OR 5.29, 95% CI: 2.09-13.43) were associated with greater likelihood of follow-up during the pandemic. Discussion: Follow-up adherence remains limited after teleophthalmology screening even in a highly insured patient population, with a further decline observed during the COVID-19 pandemic. Our results suggest that rural patients and those with lower socioeconomic status experienced greater barriers to follow-up eye care during the COVID-19 pandemic. Conclusions: Addressing barriers to in-person follow-up care is needed to effectively improve clinical outcomes after teleophthalmology screening.

Ocular Emergencies During the Coronavirus Disease 'Safer at Home Order' in Wisconsin.

INTRODUCTION: The coronavirus pandemic created large shifts in utilization of hospital resources, patient presentations, and delivery of medical care. OBJECTIVES: This retrospective study evaluated the ocular emergencies at a tertiary-care academic hospital in Wisconsin during the COVID-19-related "Safer at Home" order. METHODS: Ophthalmology consultations performed March 23 through May 26, 2020, were compared to the same time period in the 4 preceding years and the subsequent year. Billing codes were obtained to evaluate the diagnoses and procedures performed during this time frame. RESULTS: In 2020, 155 consultations were performed (42 emergency department, 113 inpatient), compared to a mean of 214 over the 5 other study years. The incidence rate ratio (IRR) of total consultations in 2020 was 0.72 (P ≤ 0.001) compared to previous years. Significantly fewer emergency department consultations were performed (IRR 0.62, P ≤ 0.001), while inpatient consultations were similar (IRR 0.88, P = 0.119). The most common diagnosis across all study years was fracture of the skull/orbit with injury to the eye/orbit. In 2020, 13% of consultations led to a procedure, compared to a total of 16% in the other years (IRR 0.59, P = 0.018). CONCLUSIONS: This study demonstrated a 28% reduction in ophthalmology consultations at a major university hospital in Wisconsin during the COVID-19-related "Safer at Home" order, though the number of consultations leading to surgery were stable. This suggests that while patients with less acute needs may have deferred care, those requiring urgent surgery still presented to the emergency department. These data may help hospitals appropriately allocate eye care resources during future public health emergencies.

Detection of Stroke with Retinal Microvascular Density and Self-Supervised Learning Using OCT-A and Fundus Imaging.

Acute cerebral stroke is a leading cause of disability and death, which could be reduced with a prompt diagnosis during patient transportation to the hospital. A portable retina imaging system could enable this by measuring vascular information and blood perfusion in the retina and, due to the homology between retinal and cerebral vessels, infer if a cerebral stroke is underway. However, the feasibility of this strategy, the imaging features, and retina imaging modalities to do this are not clear. In this work, we show initial evidence of the feasibility of this approach by training machine learning models using feature engineering and self-supervised learning retina features extracted from OCT-A and fundus images to classify controls and acute stroke patients. Models based on macular microvasculature density features achieved an area under the receiver operating characteristic curve (AUC) of 0.87-0.88. Self-supervised deep learning models were able to generate features resulting in AUCs ranging from 0.66 to 0.81. While further work is needed for the final proof for a diagnostic system, these results indicate that microvasculature density features from OCT-A images have the potential to be used to diagnose acute cerebral stroke from the retina.

Implementation of a Large-Scale Image Curation Workflow Using Deep Learning Framework.

Purpose: The curation of images using human resources is time intensive but an essential step for developing artificial intelligence (AI) algorithms. Our goal was to develop and implement an AI algorithm for image curation in a high-volume setting. We also explored AI tools that will assist in deploying a tiered approach, in which the AI model labels images and flags potential mislabels for human review. Design: Implementation of an AI algorithm. Participants: Seven-field stereoscopic images from multiple clinical trials. Methods: The 7-field stereoscopic image protocol includes 7 pairs of images from various parts of the central retina along with images of the anterior part of the eye. All images were labeled for field number by reading center graders. The model output included classification of the retinal images into 8 field numbers. Probability scores (0-1) were generated to identify misclassified images, with 1 indicating a high probability of a correct label. Main Outcome Measures: Agreement of AI prediction with grader classification of field number and the use of probability scores to identify mislabeled images. Results: The AI model was trained and validated on 17 529 images and tested on 3004 images. The pooled agreement of field numbers between grader classification and the AI model was 88.3% (kappa, 0.87). The pooled mean probability score was 0.97 (standard deviation [SD], 0.08) for images for which the graders agreed with the AI-generated labels and 0.77 (SD, 0.19) for images for which the graders disagreed with the AI-generated labels (P < 0.0001). Using receiver operating characteristic curves, a probability score of 0.99 was identified as a cutoff for distinguishing mislabeled images. A tiered workflow using a probability score of < 0.99 as a cutoff would include 27.6% of the 3004 images for human review and reduce the error rate from 11.7% to 1.5%. Conclusions: The implementation of AI algorithms requires measures in addition to model validation. Tools to flag potential errors in the labels generated by AI models will reduce inaccuracies, increase trust in the system, and provide data for continuous model development.