The Solution for the CVPR2023 NICE Image Captioning Challenge (preprint)

In this paper, we present our solution to the New frontiers for Zero-shot Image Captioning Challenge. Different from the traditional image captioning datasets, this challenge includes a larger new variety of visual concepts from many domains (such as COVID-19) as well as various image types (photographs, illustrations, graphics). For the data level, we collect external training data from Laion-5B, a large-scale CLIP-filtered image-text dataset. For the model level, we use OFA, a large-scale visual-language pre-training model based on handcrafted templates, to perform the image captioning task. In addition, we introduce contrastive learning to align image-text pairs to learn new visual concepts in the pre-training stage. Then, we propose a similarity-bucket strategy and incorporate this strategy into the template to force the model to generate higher quality and more matching captions. Finally, by retrieval-augmented strategy, we construct a content-rich template, containing the most relevant top-k captions from other image-text pairs, to guide the model in generating semantic-rich captions. Our method ranks first on the leaderboard, achieving 105.17 and 325.72 Cider-Score in the validation and test phase, respectively.

Intra-host dynamic variations in SARS-CoV-2 (preprint)

The mutations make uncertain to SARS-CoV-2 disease control and vaccine development. At population-level, single nucleotide polymorphism (SNPs) have displayed mutations for illustrating epidemiology, transmission, and pathogenesis of COVID-19. These mutations are to be expected by the analysis of intra-host level, which presented as intra-host variations (iSNVs). Here, we performed spatio-temporal analysis on iSNVs in 402 clinical samples from 170 patients, and observed an increase of genetic diversity along the day post symptom onset within individual patient and among subpopulations divided by gender, age, illness severity and viral shedding time, suggested a positive selection at intra-host level. The comparison of iSNVs and SNPs displayed that most of nonsynonymous mutations were not fixed suggested a purifying selection. This two-step fitness selection enforced iSNVs containing more nonsynonymous mutations, that highlight the potential characters of SARS-CoV-2 for viral infections and global transmissions.

Decrease of exercise endurance in critically ill COVID-19 survivors: 4 case reports (preprint)

Background: The Coronavirus Disease 2019 (COVID-19) already have been as a pandemic. However, knowledge about the sequelae of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection remains limited. Here we descirbe the pulmonary function test (PFT) and cardiopulmonary exercise test (CPET) of critically ill COVID-19 in four cases with sereve acute respiratory distress syndrome (ARDS) after discharge.Case presentation: We introduce four patients who complained of fever, cough, chest tightness and other symptoms, all of them were confirmed as SARS-CoV-2 infection by real-time reverse transcription polymerase chain reaction (RT-PCR). They were treated with mechanical ventilation because of severe ARDS. After respiratory support, antiviral and anti-infective treatment, they were weaned from mechanic ventilation with the improvement of hypoxemia. All patients were discharged from the hospital after completion of treatment and had no mortality. Around 1-month post-discharge, they were followed up for chest computed tomography (CT) scan, and performed PFT and CPET. Peak oxygen uptake of predicted (peakVO2% pred) decreased in all four cases, although spirometry were in the normal range, and only 2 cases had mild decline in carbon monoxide diffusion capacity of predicted (DLCO%pred).Conclusions: We found reduced exercise endurance in all four COVID-19 survivors, even parts of them with normal or slightly abnormal static lung function. We also believe that exercise endurance impairment of COVID-19 convalescents is more likely affected by extrapulmonary factors. Taken the above into consideration, our study highlights that the combination of PFT and CPET are important tests for tracking the development and recovery of COVID-19 survivors.