Deep Learning Mask Face Recognition with Annealing Mechanism

Face recognition (FR) has matured with deep learning, but due to the COVID-19 epidemic, people need to wear masks outside to reduce the risk of infection, making FR a challenge. This study uses the FaceNet approach combined with transfer learning using three different sizes of validated CNN architectures: InceptionResNetV2, InceptionV3, and MobileNetV2. With the addition of the cosine annealing (CA) mechanism, the optimizer can automatically adjust the learning rate (LR) during the model training process to improve the efficiency of the model in finding the best solution in the global domain. The mask face recognition (MFR) method is accomplished without increasing the computational complexity using existing methods. Experimentally, the three models of different sizes using the CA mechanism have a better performance than the fixed LR, step and exponential methods. The accuracy of the three models of different sizes using the CA mechanism can reach a practical level at about 93%.

Localized and Whole-Room Effects of Portable Air Filtration Units on Aerosol Particle Deposition and Concentration in a Classroom Environment

In indoor environments with limited ventilation, recirculating portable air filtration (PAF) units may reduce COVID-19 infection risk via not only the direct aerosol route (i.e., inhalation) but also via an indirect aerosol route (i.e., contact with the surface where aerosol particles deposited). We systematically investigated the impact of PAF units in a mock classroom, as a supplement to background ventilation, on localized and whole-room surface deposition and particle concentration. Fluorescently tagged particles with a volumetric mean diameter near 2 mu m were continuously introduced into the classroom environment via a breathing simulator with a prescribed inhalation-exhalation waveform. Deposition velocities were inferred on >50 horizontal and vertical surfaces throughout the classroom, while aerosol concentrations were spatially monitored via optical particle spectrometry. Results revealed a particle decay rate consistent with expectations based upon the reported dean air delivery rates of the PAP units. Additionally, the PAP units reduced peak concentrations by a factor of around 2.5 compared to the highest concentrations observed and led to a statistically significant reduction in deposition velocities for horizontal surfaces >2.5 m from the aerosol source. Our results not only confirm that PAF units can reduce particle concentrations but also demonstrate that they may lead to reduced particle deposition throughout an indoor environment when properly positioned with respect to the location of the particle source(s) within the room (e.g., where the largest group of students sit) and the predominant air distribution profile of the room.

[The efficacy and safety of lopinavir/ritonavir and arbidol in patients with coronavirus disease 2019].

Objective: To evaluate the efficacy and safety of lopinavir/ritonavir (LPV/r) and arbidol in treating patients with coronavirus disease 2019 (COVID-19) in the real world. Methods: The clinical data of 178 patients diagnosed with COVID-19 admitted to Guangzhou Eighth People's Hospital from January 20 to February 10, 2020 were retrospectively analyzed. According to patient's antiviral treatment regimens, 178 patients were divided into 4 groups including LPV/r group (59 patients), arbidol group (36 patients), LPV/r plus arbidol combination group (25 patients) and the supportive care group without any antiviral treatment (58 patients). The primary end point was the negative conversion time of nucleic acid of 2019 novel coronavirus (2019-nCoV) by pharyngeal swab. Results: The baseline parameters of 4 groups before treatment was comparable. The negative conversion time of viral nucleic acid was (10.20±3.49), (10.11±4.68), (10.86±4.74), (8.44±3.51) days in LPV/r group, arbidol group, combination group, and supportive care group respectively (F=2.556, P=0.058). There was also no significant difference in negative conversion rate of 2019-nCoV nucleic acid, the improvement of clinical symptoms, and the improvement of pulmonary infections by CT scan (P>0.05). However, a statistically significant difference was found in the changing rates from mild/moderate to severe/critical type at day 7 (χ(2)=9.311, P=0.017), which were 24%(6/25) in combination group, 16.7%(6/36) in arbidol group, 5.4%(3/56) in LPV/r group and 5.2%(3/58) in supportive care group. Moreover, the incidence of adverse reactions in three antiviral groups was significantly higher than that in supportive care group (χ(2)=14.875, P=0.002). Conclusions: Antiviral treatment including LPV/r or arbidol or combination does not shorten the negative conversion time of 2019-nCoV nucleic acid nor improve clinical symptoms. Moreover, these antiviral drugs cause more adverse reactions which should be paid careful attention during the treatment.

The clinical warning index of severe COVID-19: discussion based on the clinical features and outcomes of 4 patients.

OBJECTIVE: Coronavirus disease 2019 (COVID-19) is spreading worldwide. The onset of severe COVID-19 could lead to multiple organ damage and even death. It is worth paying attention to the warning index of the onset for severe COVID-19 so that patients can be identified and monitored carefully. PATIENTS AND METHODS: The report is a retrospective study that describes and analyzes the clinical features during the treatment of COVID-19. Four patients with COVID-19 were involved in this study, who were father-and-son pairs from two families. All patients were treated with the same combination of anti-microbial and anti-viral agents for 10-14 days, adjusting for the disease status. The primary outcome measure was SARS-CoV-2 detection using RT-qPCR with oropharyngeal swabs. Chest CT imaging served as a secondary outcome measure. RESULTS: One of the four patients progressed to severe disease, while the remaining patients recovered with the same treatment. A persistent decrease in the lymphocyte ratio and increase in the C-reactive protein (CRP) level were observed in the severe patient, along with other typical symptoms of COVID-19. CONCLUSIONS: The cases we described indicate that blood cell and CRP tests could be useful risk warnings of severe onset of COVID-19.

[Treatment strategies of Budd-Chiari syndrome during the epidemic period of 2019 coronavirus disease].

Prevention and control about the situation of 2019 coronavirus disease (COVID-19) are grim at present. In addition to supporting the frontline actively, medical workers in general surgery spare no efforts in making good diagnosis and treatment of specialized diseases by optimizing treatment process, providing medical advice online, mastering indications of delayed operation and emergency operation reasonably, etc. Budd-Chiari syndrome is a complex disorder, and severity of the disease varies, serious cases can be life threatening. While fighting the epidemic, medical workers should also ensure the medical needs of patients. However, instead of continuing the traditional treatment, a new management system should be developed. Based on the characteristics of Budd-Chiari syndrome patients in China and our experience, patients are divided into ordinary cases and critical cases, and treatment strategies suitable for the epidemic period of COVID-19 are put forward for reference and discussion by physicians.