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 Participants In 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 dataset. 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 datasets, 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 dataset 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 datasets 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.

To examine the associations of near work related parameters with spherical equivalent refraction and axial length in Chinese children. A total of 1770 grade 7 students with mean age of 12.7 years were examined with cycloplegic autorefraction and axial length. Questions were asked regarding time spent in near work and outdoors per day, and near work related parameters. Multivariate models revealed the following associations with greater odds of myopia: continuous reading (> 45min), odds ratio [OR], 1.4; 95% confidence interval [CI], 1.1-1.8; close television viewing distance (≤ 3m), OR, 1.7; 95% CI, 1.2-2.3; head tilt when writing, OR, 1.3; 95% CI, 1.1-1.7, and desk lighting using fluorescent vs. incandescent lamp, OR, 1.5; 95% CI, 1.2-2.0. These factors, together with close reading distance and close nib-to-fingertip distance were significantly associated with greater myopia (P<0.01). Among near work activities, only reading more books for pleasure was significantly associated with greater myopia (P=0.03). Television viewing distance (≤ 3 m), fluorescent desk light, close reading distance (≤20 cm) and close nib-to-fingertip distance (≤ 2 cm) were significantly associated with longer axial length (P<0.01). Reading distance, desk light, and reading books for pleasure had significant interaction effects with parental myopia. Continuous reading, close distances of reading, television viewing and nib-to-fingertip, head tilt when writing, reading more books for pleasure and use of fluorescent desk light were significantly associated with myopia in 12-year-old Chinese children, which indicates that visual behaviors and environments may be important factors mediating the effects of near work on myopia.

Myopia has become a critical public health problem among both children and adults, especially in some East and Southeast Asian countries such as China and Singapore.^1,2 A 2016 review³ predicted that approximately half of the world’s population will have myopia by 2050, with 10% being high myopia.³ In China, the prevalence of myopia reaches 5.2% in children aged 6 years, nearly 70% in seventh-grade students, and exceeds 80% in university students.^4-6 Myopia is not only the most common cause of avoidable visual impairment and blindness, but high or pathologic myopia is also associated with increased risk of irreversible blinding conditions, including myopic retinopathy, retinal detachment, choroidal neovascularization, and glaucoma,^7-10 leading to a heavy cost burden on individuals and communities.¹¹

Purpose: To report the design, methods and baseline data of the Anyang Childhood Eye Study (ACES), aiming to determine the prevalence, incidence and risk factors of myopia and other ocular diseases in children in central China. Methods: The ACES was a school-based cohort study conducted in Anyang city. Students have been examined and will be followed-up annually for 3–5 years. Ocular examinations included visual acuity, identification of amblyopia and strabismus, ocular biometry, optical coherence tomography, retinal photography, cycloplegic autorefraction, accommodative response, peripheral refraction, visual perception and so on. A questionnaire survey was performed to collect risk factors for myopia including near work, outdoor activity, parental myopia, birth history, habits of wearing spectacles and reading, living habits, food habits and so on. Results: Of 3112 grade 1 and 2363 grade 7 students eligible for the ACES, 2893 (93.0%) and 2267 (95.9%) participated in the study, with a mean age of 7.1 years (range 5.7–9.3 years) and 12.7 years (range 10.0–15.9 years), and proportions of male to be 57.8% and 50.0%, respectively. The prevalence rates of myopia, high myopia, emmetropia and hyperopia were 3.9%, 0.1%, 72.9% and 23.3% in grade 1, and 67.3%, 2.7%, 31.4% and 1.2% in grade 7, respectively. Similar data were found in the children confined in 6-year-old and 12-year-old ages. Conclusions: The ACES is the first large-scale cohort study in China with baseline response rates over 90%. Continuous documentation of changes and risk factors of refractive errors in these cohorts would provide new insights into myopia control in school-aged children.