Retinal fundus diseases can lead to irreversible visual impairment without timely diagnoses and appropriate treatments. Single disease-based deep learning algorithms had been developed for the detection of diabetic retinopathy, age-related macular degeneration, and glaucoma. Here, we developed a deep learning platform (DLP) capable of detecting multiple common referable fundus diseases and conditions (39 classes) by using 249,620 fundus images marked with 275,543 labels from heterogenous sources. Our DLP achieved a frequency-weighted average F1 score of 0.923, sensitivity of 0.978, specificity of 0.996 and area under the receiver operating characteristic curve (AUC) of 0.9984 for multi-label classification in the primary test dataset and reached the average level of retina specialists. External multihospital test, public data test and tele-reading application also showed high efficiency for multiple retinal diseases and conditions detection. These results indicate that our DLP can be applied for retinal fundus disease triage, especially in remote areas around the world.

Purpose To determine the prevalence and risk factors for diabetic retinopathy (DR) and sight-threatening diabetic retinopathy (STDR) in a multi-hospital-based DR screening programme among patients with diabetes in China, the Lifeline Express Diabetic Retinopathy Screening Program. Methods Patients with diabetes in eight hospitals across mainland China (both southern and northern) from January 2014 to July 2016 were recruited in this programme. All participants underwent a standardised interview and examinations and received digital fundus photography. DR severity was graded from retinal fundus photographs by retina specialists in the reading centre of Joint Shantou International Eye Center, according to the grading standards of the English National Screening Programme. STDR was defined as the presence of preproliferative DR (R2), proliferative DR (R3) and/or maculopathy (M1). Results 16 305 patients with diabetes were screened for DR in total. Fundus photographs were gradable for 15 078 patients (92.5%). The age–gender-standardised prevalence of any DR and STDR was 27.9% (95% CI, 27.2% to 28.6%) and 12.6% (95% CI, 12.1% to 13.1%), respectively. In the multiple logistic regression analysis, younger age (OR, 0.967), longer duration of diabetes (OR, 1.093), higher haemoglobin A1c (OR, 1.115), higher fasting plasma glucose (OR, 1.074), higher systolic blood pressure (OR, 1.014), faster heart rate (OR, 1.010), higher low-density lipoprotein (OR, 1.149), lower triglycerides (OR, 0.975), higher blood urea nitrogen (BUN) (OR, 1.012) and elevated serum creatinine level (OR, 1.003) were associated with the presence of DR. Similar risk factors, except for BUN and triglycerides, were found for STDR. Conclusions The prevalence of DR and STDR in diabetes was 27.9% and 12.6%, respectively in this multi-hospital-based population across China. Compared with Western population with diabetes, similar risk factors for DR and STDR were found in Chinese patients with diabetes.

Uncorrected refractive error is the leading cause of visual disability among school-aged children of European,^1,2South Asian,³and East Asian^4,5descent. As a result, programs to improve children's access to glasses have become increasingly popular.^6-8Recent reports have demonstrated, however, that rates of spectacles wearing may be low, even among children who have recently been provided free glasses.⁷Accuracy of the prescription in currently worn spectacles also appears to be an important problem in at least some parts of the world.⁹

Failure to recognize samples from the classes unseen during training is a major limitation of artificial intelligence in the real-world implementation for recognition and classification of retinal anomalies. We establish an uncertainty-inspired open set (UIOS) model, which is trained with fundus images of 9 retinal conditions. Besides assessing the probability of each category, UIOS also calculates an uncertainty score to express its confidence. Our UIOS model with thresholding strategy achieves an F1 score of 99.55%, 97.01% and 91.91% for the internal testing set, external target categories (TC)-JSIEC dataset and TC-unseen testing set, respectively, compared to the F1 score of 92.20%, 80.69% and 64.74% by the standard AI model. Furthermore, UIOS correctly predicts high uncertainty scores, which would prompt the need for a manual check in the datasets of non-target categories retinal diseases, low-quality fundus images, and non-fundus images. UIOS provides a robust method for real-world screening of retinal anomalies.

Background: The global rising prevalence and incidence of multiple sclerosis (MS) has been reported during the past decades. However, details regarding the evolution of MS burden have not been fully studied. This study aimed to investigate the global, regional, and national burden and temporal trends in MS incidence, deaths, and disability-adjusted life years (DALYs) from 1990 to 2019 using the age-period-cohort analysis.Methods: We performed a secondary comprehensive analysis of incidence, deaths, and DALYs of MS by calculating the estimated annual percentage change from 1990 to 2019 obtained from the Global Burden of Disease (GBD) 2019 study. The independent age, period, and birth cohort effects were evaluated by an age-period-cohort model.Results: In 2019, there were 59,345 incident MS cases and 22,439 MS deaths worldwide. The global number of incidences, deaths, and DALYs of MS followed an upward trend, whereas the age-standardized rates (ASR) slightly declined from 1990 to 2019. High socio-demographic index (SDI) regions had the highest ASR of incidences, deaths, and DALYs in 2019, while the rate of deaths and DALYs in medium SDI regions are the lowest. Six regions which include high-income North America, Western Europe, Australasia, Central Europe, and Eastern Europe had higher ASR of incidences, deaths, and DALYs than other regions in 2019. The age effect showed that the relative risks (RRs) of incidence and DALYs reached the peak at ages 30–39 and 50–59, respectively. The period effect showed that the RRs of deaths and DALYs increased with the period. The cohort effect showed that the later cohort has lower RRs of deaths and DALYs than the early cohort.Conclusion: The global cases of incidence, deaths, and DALYs of MS have all increased, whereas ASR has declined, with different trends in different regions. High SDI regions such as European countries have a substantial burden of MS. There are significant age effects for incidence, deaths, and DALYs of MS globally, and period effects and cohort effects for deaths and DALYs.

Retinopathy of prematurity (ROP) is a leading cause of visual impairment and irreversible blindness of children worldwide, mainly affecting preterm infants with extremely low birth weight and those who are small for gestational age. Approximately 1.2% (184 700 of 1 4900 000) of preterm infants worldwide have been estimated to have ROP, among which 30 000 preterm infants have permanent visual impairment.¹ Poor visual outcomes from ROP can be largely avoided if ROP is detected early and treated appropriately.²