Purpose: To compare the prevalence of computer vision syndrome in university students of different teaching modes during the SARS-CoV-2 outbreak period.Methods: A cross-sectional survey study using the validated Computer Vision Syndrome Questionnaire in Chinese medical students of Sichuan University who took classroom lectures and the same-grade foreign students from a Bachelor of Medicine and Bachelor of Surgery (MBBS) program who took online lectures with similar schedules.Results: A total of 137 responses from 63 Chinese students and 74 MBBS students were obtained. The highest frequency of digital screen time was 7-9 h (43.24%, 32/74) for MBBS students and 2-4 h (46.03%, 29/63) for Chinese students. The prevalence of computer vision syndrome among Chinese students and MBBS students were 50.79% and 74.32%, respectively (P = 0.004). The average numbers of reported symptoms were 5.00 ± 2.17 in Chinese students and 5.91 ± 1.90 in MBBS students (P = 0.01). The three most highly reported symptoms were “heavy eyelids” (53.97%), “dryness” (50.79%), and “feeling of a foreign body” (46.03%) in Chinese students and “dryness” (72.97%), “feeling of a foreign body” (62.16%), and “heavy eyelids” (58.11%) in MBBS students. The sum grades of computer vision syndrome had a moderate positive correlation with screen time (Spearman's correlation coefficient = 0.386, P < 0.001). The grades of symptoms of “feeling of a foreign body,” “heavy eyelids,” and “dryness” showed a weak positive correlation with screen time (Spearman's correlation coefficients were 0.220, 0.205, and 0.230, respectively).Conclusion: Online study may contribute to the prevalence of computer vision syndrome among university students.

Objectives To investigate dry eye disease (DED) in patients affected by systemic lupus erythematosus (SLE). Methods We conducted a systematic search of the literature on PubMed, EMBASE and the Cochrane Library databases from conception to 30 April 2020 for studies related to dry eye, secondary Sjögren’s syndrome (sSS) and SLE. Original full-text articles with the number of patients diagnosed with SLE of over 15 were included. The risk of bias was evaluated with a validated critical appraisal tool which assessed study quality based on confounding factors, selection bias, bias related to measurement and bias related to data analysis. Data were extracted and pooled to evaluate the overall prevalence of DED with the random-effect model and sSS with the fixed effect model. Results A total of 29 studies were included and 18 273 participants were involved. The pooled data showed that the overall prevalence of DED was 16% (95% CI 10% to 21%, p<0.001) in patients of SLE. Dry eye symptoms and abnormal Schirmer’s test were found in 26% (95% CI 20% to 32%, p<0.001) and 24% (95% CI 14% to 34%, p<0.001) of patients with SLE, respectively. 12% (95% CI 9% to 15%, p<0.001) of patients also met the criteria of sSS. The OR of DED in patients with SLE was 4.26 (95% CI 3.47 to 5.05, p<0.001) compared with healthy controls. The meta-regression analysis showed that the sample size (p=0.004) and study location (p=0.022) could be the source of heterogeneity. Conclusions DED and sSS are both common in patients with SLE.

Background: Deep learning has been extensively used in digital histopathology. The purpose of this study was to test deep learning (DL) algorithms for predicting the vital status of whole-slide image (WSI) of uveal melanoma (UM). Methods: We developed a deep learning model (Google-net) to predict the vital status of UM patients from histopathological images in TCGA-UVM cohort and validated it in an internal cohort. The histopathological DL features extracted from the model and then were applied to classify UM patients into two subtypes. The differences between two subtypes in clinical outcomes, tumor mutation, and microenvironment, and probability of drug therapeutic response were investigated further. Results: We observed that the developed DL model can achieve a high accuracy of > = 90% for patches and WSIs prediction. Using 14 histopathological DL features, we successfully classified UM patients into Cluster1 and Cluster2 subtypes. Compared to Cluster2, patients in the Cluster1 subtype have a poor survival outcome, increased expression levels of immune-checkpoint genes, higher immune-infiltration of CD8 + T cell and CD4 + T cells, and more sensitivity to anti-PD-1 therapy. Besides, we established and verified prognostic histopathological DL-signature and gene-signature which outperformed the traditional clinical features. Finally, a well-performed nomogram combining the DL-signature and gene-signature was constructed to predict the mortality of UM patients. Conclusions: Our findings suggest that DL model can accurately predict vital status in UM patents just using histopathological images. We found out two subgroups based on histopathological DL features, which may in favor of immunotherapy and chemotherapy. Finally, a well-performing nomogram that combines DL-signature and gene-signature was constructed to give a more straightforward and reliable prognosis for UM patients in treatment and management.

Introduction Deep learning (DL) has been widely used to estimate clinical images. The objective of this project was to create DL models to predict the early postoperative visual acuity after small-incision lenticule extraction (SMILE) surgery. Methods We enrolled three independent patient cohorts (a retrospective cohort and two prospective SMILE cohorts) who underwent the SMILE refractive correction procedure at two different refractive surgery centers from July to September 2022. The medical records and surgical videos were collected for further analysis. Based on the uncorrected visual acuity (UCVA) at 24 h postsurgery, the eyes were divided into two groups: those showing good recovery and those showing poor recovery. We then trained a DL model (Resnet50) to predict eyes with early postoperative visual acuity of patients in the retrospective cohort who had undergone SMILE surgery from surgical videos and subsequently validated the model’s performance in the two prospective cohorts. Finally, Gradient-weighted Class Activation Mapping (Grad-CAM) was performed for interpretation of the model. Results Among the 318 eyes (159 patients) enrolled in the study, 10,176 good quality femtosecond laser scanning images were obtained from the surgical videos. We observed that the developed DL model achieved a high accuracy of 96% for image prediction. The area under the curve (AUC) value of the DL model in the retrospective cohort was 0.962 and 0.998 in the training and validation datasets, respectively. The AUC values in two prospective cohorts were 0.959 and 0.936. At the video level, the trained machine learning (ML) model (XGBoost) also accurately distinguished patients with good or poor recovery. The AUC value of the ML model was 0.998 and 0.889 in the retrospective cohort (training and test datasets, respectively) and 1.000 and 0.984 in the two prospective cohorts. We also trained a DL model which can accurately distinguish suction loss (100%), black spots (85%), and opaque bubble layer (96%). The Grad-CAM heatmap indicated that our models can recognize the area of scanning and precisely identify intraoperative complications. Conclusions Our findings suggest that artificial intelligence (DL and ML model) can accurately predict the early postoperative visual acuity and intraoperative complications after SMILE surgery just using surgical videos or images, which may display a great importance for artificial intelligence in application of refractive surgeries.

Introduction The primary objective of this study was to develop an end-to-end model that can accurately identify corneal endothelial cells and diagnose keratoconus based on corneal endothelial images acquired from a non-contact specular microscope. Methods This was a retrospective case–control study performed at the Refractive Surgery Center of West China Hospital. A total of 403 keratoconus eyes (221 patients) and 370 myopic eyes (185 normal controls) were consecutively recruited from January 2021 to September 2022. Specular microscopy was used to image and measure the morphometric parameters of the corneal endothelial cells. A Fully Convolutional Network model with a ResNet50 (FCN_ResNet50) was established to perform the endothelial segmentation. The images were then classified using an ensemble machine learning system consisting of four pre-trained deep learning networks: DenseNet121, ResNet50, Inception_v3, and MobileNet_v2. The performance of the models was evaluated based on different metrics, such as accuracy, intersection over union (IoU), and mean IoU. Results We established a fully end-to-end deep-learning model for the segmentation of endothelial and diagnosis of keratoconus. For endothelial segmentation, the accuracy of the FCN_ResNet50 model achieved near 90% with mean IoU converging to about 80%. The ensemble machine learning system can achieve over 92% accuracy, and > 98% area under curve (AUC) values to diagnose keratoconus with endothelial cell images. In addition, we constructed a diagnostic model based on deep-learning features and developed an associated nomogram which manifested an excellent performance for diagnosis and monitoring the progression of keratoconus. Conclusions Our research developed an end-to-end model to automatically identify and assess corneal endothelial morphological changes in keratoconus eyes. Moreover, we also constructed a novel nomogram, which can provide valuable information for the diagnosis, monitoring, and management of the disease.