Risk Stratification for Diabetic Retinopathy Screening Order Using Deep Learning: A Multicenter Prospective Study.

Purpose: Real-world evaluation of a deep learning model that prioritizes patients based on risk of progression to moderate or worse (MOD+) diabetic retinopathy (DR). Methods: This nonrandomized, single-arm, prospective, interventional study included patients attending DR screening at four centers across Thailand from September 2019 to January 2020, with mild or no DR. Fundus photographs were input into the model, and patients were scheduled for their subsequent screening from September 2020 to January 2021 in order of predicted risk. Evaluation focused on model sensitivity, defined as correctly ranking patients that developed MOD+ within the first 50% of subsequent screens. Results: We analyzed 1,757 patients, of which 52 (3.0%) developed MOD+. Using the model-proposed order, the model's sensitivity was 90.4%. Both the model-proposed order and mild/no DR plus HbA1c had significantly higher sensitivity than the random order (P < 0.001). Excluding one major (rural) site that had practical implementation challenges, the remaining sites included 567 patients and 15 (2.6%) developed MOD+. Here, the model-proposed order achieved 86.7% versus 73.3% for the ranking that used DR grade and hemoglobin A1c. Conclusions: The model can help prioritize follow-up visits for the largest subgroups of DR patients (those with no or mild DR). Further research is needed to evaluate the impact on clinical management and outcomes. Translational Relevance: Deep learning demonstrated potential for risk stratification in DR screening. However, real-world practicalities must be resolved to fully realize the benefit.

Patient, family member, and health care provider perspective on barriers and facilitators to diabetic retinopathy screening in Thailand: A qualitative study.

OBJECTIVES: Diabetic retinopathy (DR) can cause significant visual impairment which can be largely avoided by early detection through proper screening and treatment. People with DR face a number of challenges from early detection to treatment. The aim of this study was to investigate factors that influence DR screening in Thailand and to identify barriers to follow-up compliance from patient, family member, and health care provider (HCP) perspectives. METHODS: A total of 15 focus group discussions (FGDs) were held, each with five to twelve participants. There were three distinct stakeholders: diabetic patients (n = 47) presenting to a diabetic retinopathy clinic in Thailand, their family members (n = 41), and health care providers (n = 34). All focus group conversations were transcribed verbatim. Thematic analysis was used to examine textual material. RESULTS: Different themes emerged from the FGD on knowledge about diabetes, self-care behaviors of diabetes mellitus (DM), awareness about DR, barriers to DR screening, and the suggested solutions to address those barriers. Data showed lower knowledge and awareness about diabetes and DR in both patients and family members. Long waiting times, financial issues, and lack of a person to accompany appointments were identified as the major deterrents for attending DR screening. Family support for patients was found to vary widely, with some patients reporting to have received adequate support while others reported having received minimal support. Even though insurance covered the cost of attending diabetes/DR screening program, some patients did not show up for their appointments. CONCLUSION: Patients need to be well-informed about the asymptomatic nature of diabetes and DR. Communication at the patient level and shared decision-making with HCPs are essential. Family members and non-physician clinicians (such as diabetes nurses, diabetes educators, physician assistants) who work in the field of diabetes play a vital role in encouraging patients to attend diabetes and DR follow-ups visits regularly.

Diagnostic performance of anterior segment optical coherence tomography in detecting plateau iris.

Objective: To evaluate the diagnostic performance of manual grading of anterior segment optical coherence tomography (AS-OCT) in detection of plateau iris configuration (PIC) based on the presence of standardised ultrasound biomicroscopy (UBM) criteria in at least two quadrants; namely, clinical diagnosis of PIC (DxPIC). Methods and analysis: In this cross-sectional study, paired AS-OCT and UBM images were evaluated by three glaucoma specialists. AS-OCT was classified into two mechanisms, PIC versus non-PIC, of primary angle closure disease (PACD) and AS-OCT-PIC diagnostic performance was tested with DxPIC. Results: One hundred and seventy-nine eyes of 142 patients were enrolled for analysis, and DxPIC was found in 85 eyes (47.49%). Intraobserver agreement rates of AS-OCT classification by the graders were 0.77, 0.701 and 0.742 (all p<0.001), and interobserver agreement rates, between a senior glaucoma specialist and the other two glaucoma specialists, were 0.68 and 0.702 (all p<0.001). Plateau iris was classified in AS-OCT images by the three graders, rated 32.96%-39.1% and 24.58%-34.08% in the horizontal and vertical axes, respectively. Diagnostic performance was analysed, yielding sensitivity ranging from 56.47% to 77.78%, and specificity of 48.94% to 64.29%. We applied disease prevalence of 30%, revealing positive predictive values varying from 32.16% to 44.44%, and negative predictive values of 72.4% to 85.71%. Accuracy ranged from 51.2% to 65%. Agreement between the two devices was fair, kappa range 0.31-0.351. Conclusion: Performance of manual grading of AS-OCT in detection of DxPIC was relatively poor; therefore, unadjusted AS-OCT does not appear to be good for manual PIC screening in PACD patients and cannot serve as a substitute for UBM in PIC detection.

Deep Learning for Anterior Segment Optical Coherence Tomography to Predict the Presence of Plateau Iris.

Purpose: The purpose of this study was to evaluate the diagnostic performance of deep learning (DL) anterior segment optical coherence tomography (AS-OCT) as a plateau iris prediction model. Design: We used a cross-sectional study of the development and validation of the DL system. Methods: We conducted a collaboration between a referral eye center and an informative technology department. The study enrolled 179 eyes from 142 patients with primary angle closure disease (PACD). All patients had remaining appositional angle after iridotomy. Each eye was scanned in four quadrants for both AS-OCT and ultrasound biomicroscopy (UBM). A DL algorithm for plateau iris prediction of AS-OCT was developed from training datasets and was validated in test sets. Sensitivity, specificity, and area under the receiver operating characteristics curve (AUC-ROC) of the DL for predicting plateau iris were evaluated, using UBM as a reference standard. Results: Total paired images of AS-OCT and UBM were from 716 quadrants. Plateau iris was observed with UBM in 276 (38.5%) quadrants. Trainings dataset with data augmentation were used to develop an algorithm from 2500 images, and the test set was validated from 160 images. AUC-ROC was 0.95 (95% confidence interval [CI] = 0.91 to 0.99), sensitivity was 87.9%, and specificity was 97.6%. Conclusions: DL revealed a high performance in predicting plateau iris on the noncontact AS-OCT images. Translational Relevance: This work could potentially assist clinicians in more practically detecting this nonpupillary block mechanism of PACD.

Primary Prophylaxis for Cryptococcosis With Fluconazole in Human Immunodeficiency Virus-Infected Patients With CD4 T-Cell Counts <100 Cells/µL and Receiving Antiretroviral Therapy.

A prospective observational cohort study was conducted in 302 human immunodeficiency virus-infected patients who had a CD4 T-cell count <100 cells/µL and negative serum cryptococcal antigen initiating antiretroviral therapy in a resource-limited setting. During 2-year follow-up, there were no differences of survival rates and occurrences of newly diagnosed cryptococcosis between patients with and without fluconazole for primary prophylaxis of cryptococcosis.