In this evaluation of retinal images from multiethnic cohorts of patients with diabetes, the DLS had high sensitivity and specificity for identifying diabetic retinopathy and related eye diseases. Further research is necessary to evaluate the applicability of the DLS in health care settings and the utility of the DLS to improve vision outcomes.

IMPORTANCE A deep learning system (DLS) that could automatically detect glaucomatous optic neuropathy (GON) with high sensitivity and specificity could expedite screening for GON.OBJECTIVE To establish a DLS for detection of GON using retinal fundus images and glaucoma diagnosis with convoluted neural networks (GD-CNN) that has the ability to be generalized across populations. DESIGN, SETTING, AND PARTICIPANTSIn this cross-sectional study, a DLS for the classification of GON was developed for automated classification of GON using retinal fundus images obtained from the Chinese Glaucoma Study Alliance, the Handan Eye Study, and online databases. The researchers selected 241 032 images were selected as the training data set. The images were entered into the databases on June 9, 2009, obtained on July 11, 2018, and analyses were performed on December 15, 2018. The generalization of the DLS was tested in several validation data sets, which allowed assessment of the DLS in a clinical setting without exclusions, testing against variable image quality based on fundus photographs obtained from websites, evaluation in a population-based study that reflects a natural distribution of patients with glaucoma within the cohort and an additive data set that has a diverse ethnic distribution. An online learning system was established to transfer the trained and validated DLS to generalize the results with fundus images from new sources. To better understand the DLS decision-making process, a prediction visualization test was performed that identified regions of the fundus images utilized by the DLS for diagnosis. EXPOSURES Use of a deep learning system. MAIN OUTCOMES AND MEASURES Area under the receiver operating characteristics curve (AUC), sensitivity and specificity for DLS with reference to professional graders. RESULTS From a total of 274 413 fundus images initially obtained from CGSA, 269 601 images passed initial image quality review and were graded for GON. A total of 241 032 images (definite GON 29 865 [12.4%], probable GON 11 046 [4.6%], unlikely GON 200 121 [83%]) from 68 013 patients were selected using random sampling to train the GD-CNN model. Validation and evaluation of the GD-CNN model was assessed using the remaining 28 569 images from CGSA. The AUC of the GD-CNN model in primary local validation data sets was 0.996 (95% CI, 0.995-0.998), with sensitivity of 96.2% and specificity of 97.7%. The most common reason for both false-negative and false-positive grading by GD-CNN (51 of 119 [46.3%] and 191 of 588 [32.3%]) and manual grading (50 of 113 [44.2%] and 183 of 538 [34.0%]) was pathologic or high myopia.CONCLUSIONS AND RELEVANCE Application of GD-CNN to fundus images from different settings and varying image quality demonstrated a high sensitivity, specificity, and generalizability for detecting GON. These findings suggest that automated DLS could enhance current screening programs in a cost-effective and time-efficient manner.

Background Screening for chronic kidney disease is a challenge in community and primary care settings, even in high-income countries. We developed an artificial intelligence deep learning algorithm (DLA) to detect chronic kidney disease from retinal images, which could add to existing chronic kidney disease screening strategies. MethodsWe used data from three population-based, multiethnic, cross-sectional studies in Singapore and China. The Singapore Epidemiology of Eye Diseases study (SEED, patients aged ≥40 years) was used to develop (5188 patients) and validate (1297 patients) the DLA. External testing was done on two independent datasets: the Singapore Prospective Study Program (SP2, 3735 patients aged ≥25 years) and the Beijing Eye Study (BES, 1538 patients aged ≥40 years). Chronic kidney disease was defined as estimated glomerular filtration rate less than 60 mL/min per 1•73m². Three models were trained: 1) image DLA; 2) risk factors (RF) including age, sex, ethnicity, diabetes, and hypertension; and 3) hybrid DLA combining image and RF. Model performances were evaluated using the area under the receiver operating characteristic curve (AUC). Findings In the SEED validation dataset, the AUC was 0•911 for image DLA (95% CI 0•886 -0•936), 0•916 for RF (0•891-0•941), and 0•938 for hybrid DLA (0•917-0•959). Corresponding estimates in the SP2 testing dataset were 0•733 for image DLA (95% CI 0•696-0•770), 0•829 for RF (0•797-0•861), and 0•810 for hybrid DLA (0•776-0•844); and in the BES testing dataset estimates were 0•835 for image DLA (0•767-0•903), 0•887 for RF (0•828-0•946), and 0•858 for hybrid DLA (0•794-0•922). AUC estimates were similar in subgroups of people with diabetes (image DLA 0•889 [95% CI 0•850-0•928], RF 0•899 [0•862-0•936], hybrid 0•925 [0•893-0•957]) and hypertension (image DLA 0•889 [95% CI 0•860-0•918], RF 0•889 [0•860-0•918], hybrid 0•918 [0•893-0•943]).Interpretation A retinal image DLA shows good performance for estimating chronic kidney disease, underlying the feasibility of using retinal photography as an adjunctive or opportunistic screening tool for chronic kidney disease in community populations.