Masked Face Recognition System Based on Attention Mechanism

With the continuous development of deep learning, the face recognition field has also developed rapidly. However, with the massive popularity of COVID-19, face recognition with masks is a problem that is now about to be tackled in practice. In recognizing a face wearing a mask, the mask obscures most of the facial features of the face, resulting in the general face recognition model only capturing part of the facial information. Therefore, existing face recognition models are usually ineffective in recognizing faces wearing masks. This article addresses this problem in the existing face recognition model and proposes an improvement of Facenet. We use ConvNeXt-T as the backbone of the network model and add the ECA (Efficient Channel Attention) mechanism. This enhances the feature extraction of the unobscured part of the face to obtain more useful information, while avoiding dimensionality reduction and not increasing the model complexity. We design new face recognition models by investigating the effects of different attention mechanisms on face mask recognition models and the effects of different data set ratios on experimental results. In addition, we construct a large set of faces wearing masks so that we can efficiently and quickly train the model. Through experiments, our model proved to be 99.76% accurate for real faces wearing masks. A combined accuracy of 99.48% for extreme environments such as too high or lousy contrast and brightness.

A More Important Role of Diabetes Mellitus in Mortality in Elderly Critical COVID-19 Patients: a Single-center Retrospective Study (preprint)

Background: COVID-19 has been circulating worldwide since December 2019. However, its independent risk factors of mortality need further insights in elderly critical COVID-19 patients. Methods: Totally 48 elderly and critically ill COVID-19 patients with clear end point were enrolled when the data were collected, with 16 discharged and 32 died. Kaplan-Meier analysis and Cox regression were performed to identify the risk factors of mortality in elderly critical COVID-19 patients. Survival curve was conducted to present the impact of Diabetes Mellitus on mortality. Mann-Whitney U and t tests were used to 39 clinical variates between the survivor and non-survivor groups. Results: As Kaplan-Meier analysis confirmed, only three variates Diabetes Mellitus, chronic obstructive pulmonary disease and family aggregation showed significant difference between the survivor and non-survivor groups. However, only variate Diabetes Mellitus presented significance in Cox regression. Higher C-reaction protein, interleukin-2 (IL-2), IL-8 and creatinine were detected in survivor group than non-survivor group, which was reverse to estimated glomerular filtration rate. The other laboratory finding showed no significant difference between survivor and non-survivor groups. Conclusions: Diabetes Mellitus, chronic obstructive pulmonary disease and family aggregation can contribute to mortality by COVID-19 while Diabetes Mellitus also reduces the survival time of elderly and critically ill COVID-19 patients, highlighting the more important role of Diabetes Mellitus. Laboratory findings could not serve as good predictors of death for elderly and critically ill COVID-19 patients.

Plasma cell-free DNA promise disease monitoring and tissue injury assessment of COVID-19 (preprint)

COVID-19 is a huge threat to global health. Due to the lack of definitive etiological therapeutics currently, effective disease monitoring is of high clinical value for better healthcare and management of the large number of COVID-19 patients. In this study, we recruited 37 COVID-19 patients, collected 176 blood samples upon diagnosis and during treatment, and analyzed cell-free DNA (cfDNA) in these samples. We report gross abnormalities in cfDNA of COVID-19 patients, including elevated GC content, altered molecule size and end motif patterns. More importantly, such cfDNA characteristics reflect patient-specific physiological conditions during treatment. Further analysis on tissue origin tracing of cfDNA reveals frequent tissue injuries in COVID-19 patients, which is supported by clinical diagnoses. Hence, we demonstrate the translational merit of cfDNA as valuable analyte for effective disease monitoring, as well as tissue injury assessment in COVID-19 patients.

Cell entry of SARS-CoV-2 conferred by angiotensin-converting enzyme 2 (ACE2) of different species (preprint)

The outbreak of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) poses a huge threat to many countries around the world. However, where is it origin and which animals are sensitive to cross-species transmission is unclear. The interaction of virus and cell receptor is a key determinant of host range for the novel coronavirus. Angiotensin-converting enzyme 2 (ACE2) is demonstrated as the primary entry receptor for SARS-CoV-2. In this study, we evaluated the SARS-CoV-2 entry mediated by ACE2 of 11 different species of animals, and discovered that ACE2 of Rhinolophus sinicus (Chinese horseshoe bat), Felis catus (domestic cat), Canis lupus familiaris (dog), Sus scrofa (pig), Capra hircus (goat) and especially Manis javanica (Malayan pangolin) were able to render SARS-CoV-2 entry in non-susceptible cells. This is the first report that ACE2 of Pangolin could mediate SARS-CoV-2 entry which increases the presume that SARS-CoV-2 may have a pangolin origin. However, none of the ACE2 proteins from Rhinolophus ferrumequinum (greater horseshoe bat), Gallus gallus (chicken), Notechis scutatus (mainland tiger snake), Mus musculus (house mouse) rendered SARS-CoV-2 entry. Specifically, a natural isoform of Macaca mulatta (Rhesus monkey) ACE2 with a mutation of Y217N was resistance to infection, which rises the possible impact of this type of ACE2 during monkey studies of SARS-CoV-2. Overall, these results clarify that SARS-CoV-2 could engage receptors of multiple species of animals and it is a perplexed work to track SARS-CoV-2 origin and its intermediate hosts. IMPORTANCEIn this study, we illustrated that SARS-CoV-2 is able to engage receptors of multiple species of animals. This indicated that it may be a perplexed work to track SARS-CoV-2 origin and discover its intermediate hosts. This feature of virus is considered to potentiate its diverse cross-species transmissibility. Of note, here is the first report that ACE2 of Pangolin could mediate SARS-CoV-2 entry which increases the possibility that SARS-CoV-2 may have a pangolin origin. And we also demonstrated that not all species of bat were sensitive to SARS-CoV-2 infection. At last, it is also important to detect the expression ratio of the Y217N ACE2 to the prototype in Rhesus monkeys to be recruited for studies on SARS-CoV-2 infection.