Background: COVID-19 has become one of the most serious global epidemics in the 21st Century. This study aims to explore the distribution of research capabilities of countries, institutions, and researchers, and the hotspots and frontiers of coronavirus research in the past two decades. In it, references for funding support of urgent projects and international cooperation among research institutions are provided. Method: the Web of Science core collection database was used to retrieve the documents related to coronavirus published from 2003 to 2020. Citespace.5.6.R2, VOSviewer1.6.12, and Excel 2016 were used for bibliometric analysis. Results: 11,036 documents were retrieved, of which China and the United States have contributed the most coronavirus studies, Hong Kong University being the top contributor. Regarding journals, the Journal of Virology has contributed the most, while in terms of researchers, Yuen Kwok Yung has made the most contributions. The proportion of documents published by international cooperation has been rising for decades. Vaccines for SARS-CoV-2 are under development, and clinical trials of several drugs are ongoing. Conclusions: international cooperation is an important way to accelerate research progress and achieve success. Developing corresponding vaccines and drugs are the current hotspots and research directions.

Diabetic retinopathy (DR) is a very common complication of diabetes that can lead to retinal damage, affecting vision. If not detected and diagnosed in time, it may result in blindness. With the development of deep learning, various automatic diagnosis models for DR have been proposed. Most models are based on convolutional neural networks, but due to the small size of convolution kernels in shallow networks, the receptive field is limited, preventing the capture of global information. This limitation forces models to require more convolutional layers to expand the receptive field and capture global information related to DR. In this paper, we use a combined module of multiscale convolution and convolutional self-attention, and we adjust the module structure, allowing the model to have both long-range dependency modeling capability and CNN’s multiscale local feature extraction capability, while having fewer model parameters than the selected comparison models. Our proposed MSAmix-Net achieved an accuracy of 82.3% in the five-class classification task on the combined dataset of APTOS-2019 and Messidor-2, and an accuracy of 93.9% in the three-class classification task on the DRAC-2022 dataset. The experiments demonstrate that the model can effectively capture and represent the complex features of diabetic retinopathy.