An RNA-hydrolyzing recombinant minibody prevents both influenza A virus and coronavirus in co-infection models.

With the lifting of COVID-19 non-pharmaceutical interventions, the resurgence of common viral respiratory infections was recorded in several countries worldwide. It facilitates viral co-infection, further burdens the already over-stretched healthcare systems. Racing to find co-infection-associated efficacy therapeutic agents need to be rapidly established. However, it has encountered numerous challenges that necessitate careful investigation. Here, we introduce a potential recombinant minibody-associated treatment, 3D8 single chain variable fragment (scFv), which has been developed as a broad-spectrum antiviral drug that acts via its nucleic acid catalytic and cell penetration abilities. In this research, we demonstrated that 3D8 scFv exerted antiviral activity simultaneously against both influenza A viruses (IAVs) and coronaviruses in three established co-infection models comprising two types of coronaviruses [beta coronavirus-human coronavirus OC43 (hCoV-OC43) and alpha coronavirus-porcine epidemic diarrhea virus (PEDV)] in Vero E6 cells, two IAVs [A/Puerto Rico/8/1934 H1N1 (H1N1/PR8) and A/X-31 (H3N2/X-31)] in MDCK cells, and a combination of coronavirus and IAV (hCoV-OC43 and adapted-H1N1) in Vero E6 cells by a statistically significant reduction in viral gene expression, proteins level, and approximately around 85%, 65%, and 80% of the progeny of 'hCoV-OC43-PEDV', 'H1N1/PR8-H3N2/X-31', and 'hCoV-OC43-adapted-H1N1', respectively, were decimated in the presence of 3D8 scFv. Taken together, we propose that 3D8 scFv is a promising broad-spectrum drug for treatment against RNA viruses in co-infection.

SLAPP: Subgraph-level attention-based performance prediction for deep learning models.

The intricacy of the Deep Learning (DL) landscape, brimming with a variety of models, applications, and platforms, poses considerable challenges for the optimal design, optimization, or selection of suitable DL models. One promising avenue to address this challenge is the development of accurate performance prediction methods. However, existing methods reveal critical limitations. Operator-level methods, proficient at predicting the performance of individual operators, often neglect broader graph features, which results in inaccuracies in full network performance predictions. On the contrary, graph-level methods excel in overall network prediction by leveraging these graph features but lack the ability to predict the performance of individual operators. To bridge these gaps, we propose SLAPP, a novel subgraph-level performance prediction method. Central to SLAPP is an innovative variant of Graph Neural Networks (GNNs) that we developed, named the Edge Aware Graph Attention Network (EAGAT). This specially designed GNN enables superior encoding of both node and edge features. Through this approach, SLAPP effectively captures both graph and operator features, thereby providing precise performance predictions for individual operators and entire networks. Moreover, we introduce a mixed loss design with dynamic weight adjustment to reconcile the predictive accuracy between individual operators and entire networks. In our experimental evaluation, SLAPP consistently outperforms traditional approaches in prediction accuracy, including the ability to handle unseen models effectively. Moreover, when compared to existing research, our method demonstrates a superior predictive performance across multiple DL models.

Efficient and accurate compound scaling for convolutional neural networks.

Designing efficient and accurate network architectures to support various workloads, from servers to edge devices, is a fundamental problem as the use of Convolutional Neural Networks (ConvNets) becomes increasingly widespread. One simple yet effective method is to scale ConvNets by systematically adjusting the dimensions of the baseline network, including width, depth, and resolution, enabling it to adapt to diverse workloads by varying its computational complexity and representation ability. However, current state-of-the-art (SOTA) scaling methods for neural network architectures overlook the inter-dimensional relationships within the network and the impact of scaling on inference speed, resulting in suboptimal trade-offs between accuracy and inference speed. To overcome those limitations, we propose a scaling method for ConvNets that utilizes dimension relationship and runtime proxy constraints to improve accuracy and inference speed. Specifically, our research notes that higher input resolutions in convolutional layers lead to redundant filters (convolutional width) due to increased similarity between information in different positions, suggesting a potential benefit in reducing filters while increasing input resolution. Based on this observation, the relationship between the width and resolution is empirically quantified in our work, enabling models with higher parametric efficiency to be prioritized through our scaling strategy. Furthermore, we introduce a novel runtime prediction model that focuses on fine-grained layer tasks with different computational properties for more accurate identification of efficient network configurations. Comprehensive experiments show that our method outperforms prior works in creating a set of models with a trade-off between accuracy and inference speed on the ImageNet datasets for various ConvNets.

Deep learning for detecting visually impaired cataracts using fundus images.

Purpose: To develop a visual function-based deep learning system (DLS) using fundus images to screen for visually impaired cataracts. Materials and methods: A total of 8,395 fundus images (5,245 subjects) with corresponding visual function parameters collected from three clinical centers were used to develop and evaluate a DLS for classifying non-cataracts, mild cataracts, and visually impaired cataracts. Three deep learning algorithms (DenseNet121, Inception V3, and ResNet50) were leveraged to train models to obtain the best one for the system. The performance of the system was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. Results: The AUC of the best algorithm (DenseNet121) on the internal test dataset and the two external test datasets were 0.998 (95% CI, 0.996-0.999) to 0.999 (95% CI, 0.998-1.000),0.938 (95% CI, 0.924-0.951) to 0.966 (95% CI, 0.946-0.983) and 0.937 (95% CI, 0.918-0.953) to 0.977 (95% CI, 0.962-0.989), respectively. In the comparison between the system and cataract specialists, better performance was observed in the system for detecting visually impaired cataracts (p < 0.05). Conclusion: Our study shows the potential of a function-focused screening tool to identify visually impaired cataracts from fundus images, enabling timely patient referral to tertiary eye hospitals.

Fisetin Prevents Angiogenesis in Diabetic Retinopathy by Downregulating VEGF.

Diabetic retinopathy (DR) is one of the more serious complications of diabetes. However, the mechanisms involved in DR are complex and still need to be investigated. The beneficial effects of fisetin have been widely reported, but whether it is beneficial in DR is not clear yet. This study was designed to investigate the regulatory role of fisetin in regulating DR and explore the involved mechanism. First, 30 mM glucose was used to establish DR cell model in vitro. Cell counting kit 8 (CCK8) assay was utilized to study the effects of fisetin on cell viability through treating human retinal microvascular endothelial cells (HRMECs) with 25, 50, and 100 µM fisetin. Transwell and wound healing assays were used to detect the function of fisetin on the migration and angiogenesis on HG-induced HRMECs. Finally, OE-VEGF was used as a mimic of VEGF, and western blotting (WB) was used to verify the targeting genes of fisetin. HG induced an increase in cell viability, cell migration, and angiogenesis in HRMECs, whereas fisetin inhibited these enhancements induced by HG through inhibiting VEGF. In conclusion, fisetin prevents angiogenesis in DR by downregulating VEGF.

Sprouty-related proteins with EVH1 domain (SPRED2) prevents high-glucose induced endothelial-mesenchymal transition and endothelial injury by suppressing MAPK activation.

Diabetic retinopathy (DR) is a common complication of diabetes, and the leading cause of blindness in adults. Sprouty-related proteins with EVH1 domain (SPRED2) play an important role in diabetes and are closely related to the lens and eye morphogenesis. This study attempted to investigate the role and related mechanism of SPRED2 in DR. DR rat model was established by administration streptozocin. Human retinal endothelial cells (HRECs) were treated with high glucose (HG) to mimic DR. The results showed that SPRED2 expression was decreased in the retinal tissues of DR rats and HG-treated HRECs. MTT assay and flow cytometry data showed that SPRED2 overexpression reduced cell viability of HG-treated HRECs. SPRED2 overexpression enhanced Caspase-3 activity and promoted apoptosis of HG-treated HRECs. Furthermore, the expressions of endothelial cell markers CD31 and E-cad were down-regulated, whereas the expressions of mesenchymal cell markers FSP1, SM22, and α-SMA were up-regulated in the HG-treated HRECs. SPRED2 overexpression reversed HG-induced endothelial-mesenchymal transition in HRECs. The expressions of tight junction components claudin 3, occludin, and ZO-1 were increased in HG-treated HRECs following SPRED2 up-regulation. In addition, SPRED2 overexpression downregulated the expression of p-ERK1/2, p-p38, and p-JNK in the HG-treated HRECs. In conclusion, this study demonstrated that SPRED2 overexpression repressed endothelial-mesenchymal transition and endothelial injury in HG-treated HRECs by suppressing MAPK signaling pathway. These findings suggested that SPRED2 may be a novel potential therapeutic target implicated in DR progression.

Group Analysis of Q Values Calculated with Tangential Radius of Curvature from Human Anterior Corneal Surface.

OBJECTIVE: To calculate the Q values from the human anterior corneal surface with the tangential radius of curvature and analyze its distribution characteristics in different age and refractive status groups. METHODS: Tangential power maps of the anterior cornea from Orbscan II were acquired for 201 subjects' right eyes. They were divided into groups of adults and children and then divided further into subgroups according to the refraction status. The Q values of each semimeridian were calculated by the tangential radius with a linear regression equation. The Q value distribution in both the nasal cornea and temporal cornea were analyzed. RESULTS: The mean temporal Q values of the emmetropia group of adults and all children's groups were significantly different from the mean nasal Q value. The mean nasal corneal Q values were more negative in children. The adult group showed differences only in the low myopia group. The mean Q value of the nasal cornea among different refractive groups of children was significantly different, and so was the temporal cornea between the adult myopia and emmetropia group. CONCLUSION: The method using the tangential radius of curvature combined with linear regression to obtain anterior surface Q values for both adults and children was stable and reliable. When we analyzed the anterior corneal Q value, area division was necessary.