Intelligent COVID-19 screening platform based on breath analysis.

The spread of coronavirus disease 2019 (COVID-19) results in an increasing incidence and mortality. The typical diagnosis technique for severe acute respiratory syndrome coronavirus 2 infection is reverse transcription polymerase chain reaction, which is relatively expensive, time-consuming, professional, and suffered from false-negative results. A reliable, non-invasive diagnosis method is in urgent need for the rapid screening of COVID-19 patients and controlling the epidemic. Here we constructed an intelligent system based on the volatile organic compound (VOC) biomarkers in human breath combined with machine learning models. The VOC profiles of 122 breath samples (65 of COVID-19 infections and 57 of controls) were identified with a portable gas chromatograph-mass spectrometer. Among them, eight VOCs exhibited significant differences (p< 0.001) between the COVID-19 and the control groups. The cross-validation algorithm optimized support vector machine (SVM) model was employed for the prediction of COVID-19 infection. The proposed SVM model performed a powerful capability in discriminating COVID-19 patients from healthy controls, with an accuracy of 97.3%, a sensitivity of 100%, a specificity of 94.1%, and a precision of 95.2%, and anF1 score of 97.6%. The SVM model was also compared with other common machine models, including artificial neural network,k-nearest neighbor, and logistic regression, and demonstrated obvious superiority in the prediction of COVID-19 infection. Furthermore, user-friendly software was developed based on the optimized SVM model. The developed intelligent platform based on breath analysis provides a new strategy for the point-of-care screening of COVID and shows great potential in clinical application.

Respiratory immune status and microbiome in recovered COVID-19 patients revealed by metatranscriptomic analyses

Coronavirus disease 2019 (COVID-19) is currently a severe threat to global public health, and the immune response to COVID-19 infection has been widely investigated. However, the immune status and microecological changes in the respiratory systems of patients with COVID-19 after recovery have rarely been considered. We selected 72 patients with severe COVID-19 infection, 57 recovered from COVID-19 infection, and 65 with non-COVID-19 pneumonia, for metatranscriptomic sequencing and bioinformatics analysis. Accordingly, the differentially expressed genes between the infected and other groups were enriched in the chemokine signaling pathway, NOD-like receptor signaling pathway, phagosome, TNF signaling pathway, NF-kappa B signaling pathway, Toll-like receptor signaling pathway, and C-type lectin receptor signaling pathway. We speculate that IL17RD, CD74, and TNFSF15 may serve as disease biomarkers in COVID-19. Additionally, principal coordinate analysis revealed significant differences between groups. In particular, frequent co-infections with the genera Streptococcus, Veillonella, Gemella, and Neisseria, among others, were found in COVID-19 patients. Moreover, the random forest prediction model with differential genes showed a mean area under the curve (AUC) of 0.77, and KCNK12, IL17RD, LOC100507412, PTPRT, MYO15A, MPDZ, FLRT2, SPEG, SERPINB3, and KNDC1 were identified as the most important genes distinguishing the infected group from the recovered group. Agrobacterium tumefaciens, Klebsiella michiganensis, Acinetobacter pittii, Bacillus sp. FJAT.14266, Brevundimonas naejangsanensis, Pseudopropionibacterium propionicum, Priestia megaterium, Dialister pneumosintes, Veillonella rodentium, and Pseudomonas protegens were selected as candidate microbial markers for monitoring the recovery of COVID patients. These results will facilitate the diagnosis, treatment, and prognosis of COVID patients recovering from severe illness.

Machine learning identifies a COVID-19-specific phenotype in university students using a mental health app (preprint)

Advances in smartphone technology have allowed people to access mental healthcare via digital apps from wherever and whenever they choose. University students experience a high burden of mental health concerns. Although these apps improve mental health symptoms, user engagement has remained low. Studies have shown that users can be subgrouped based on unique characteristics that just-in-time adaptive interventions (JITAIs) can use to improve engagement. To date, however, no studies have examined the effect of the COVID-19 pandemic on these subgroups. Here, we use machine learning to examine user subgroup characteristics across three COVID-19-specific timepoints: during lockdown, immediately following lockdown, and three months after lockdown ended. We demonstrate that there are three unique subgroups of university students who access mental health apps. Two of these, with either higher or lower mental well-being, were defined by characteristics that were stable across COVID-19 timepoints. The third, situational well-being, had characteristics that were timepoint-dependent, suggesting that they are highly influenced by traumatic stressors and stressful situations. This subgroup also showed feelings and behaviours consistent with burnout. Overall, our findings clearly suggest that user subgroups are unique: they have different characteristics and therefore likely have different mental healthcare goals. Our findings also highlight the importance of including questions and additional interventions targeting traumatic stress(ors), reason(s) for use, and burnout in JITAI-style mental health apps to improve engagement.

Comprehensive structural analysis reveals broad-spectrum neutralizing antibodies against Omicron (preprint)

The pandemic of COVID-19 caused by SARS-CoV-2 continues to spread around the world. Mutant strains of SARS-CoV-2 are constantly emerging. At present, Omicron variants have become mainstream. In this work, we carried out a systematic and comprehensive analysis of the reported spike protein antibodies, counting the antibodies' epitopes and genotypes. We further comprehensively analyzed the impact of Omicron mutations on antibody epitopes and classified these antibodies according to their binding patterns. We found that the epitopes of one class of antibodies were significantly less affected by Omicron mutations than other classes. Binding and virus neutralization experiments show that such antibodies can effectively inhibit the immune escape of Omicron. Cryo-EM results show that this class of antibodies utilizes a conserved mechanism to neutralize SARS-CoV-2. Our results greatly help us deeply understand the impact of Omicron mutations. At the same time, it also provides guidance and insights for developing Omicron antibodies and vaccines.

The prevalence and associated factors for self-harm and suicide ideation among Chinese Indonesians during the COVID-19 pandemic (preprint)

Background: This study aimed to identify the prevalence and associated factors for self-harm and suicide ideation among Chinese Indonesians during the COVID-19 pandemic. Methods: A non-random sampling was performed through a nationwide online survey in Indonesia (May-June 2021). The online survey covered participants’ demographic information, suicide literacy, suicide stigma, loneliness, and self-harm and suicide ideation. A series of t-tests, Chi-squares, and hierarchical logistic regressions with the backward-stepwise method were used to identify the factors associated with self-harm and suicide ideation. Responses from a total of 484 Chinese Indonesians were analyzed in this study. Results: The prevalence of self-harm and suicide ideation in the two weeks preceding the survey among Chinese Indonesian people was 35.5%. The predictive model showed a significant goodness-of-fit to the observed data (x2[17]=174.1, p<.001; RN2=.41). Chinese Indonesians with an average monthly income of ≥USD 843 were found to be 0.23 (95% CI=0.07-0.99) times less likely to experience self-harm and suicide ideation than those who did not have an income. A one-point increase in the intensity of suicide glorification and loneliness were associated with 3.06 and 3.67 increase in the chance of experiencing self-harm and suicide ideation, respectively. Conclusion: One-third of Chinese Indonesians self-reported self-harm and suicide ideation during the COVID-19 pandemic. Mental health and suicide prevention intervention programs are recommended to target those with low-socioeconomic status, high glorification towards suicide, and high perceived loneliness.

Bringing digital resource decision-making up to date: preparing the inaugural class of digitalised preservice teachers for informed evaluation of digital resources

The virtualised schools and universities of the Covid-19 pandemic became rely heavily on educational digital resources (EDRs), so that it has made the selection and use of high-quality EDRs even more critical for quality education. This qualitative case study aims to examine decision-making criteria used by preservice teachers (PTs) in selecting and evaluating EDRs. Twenty senior PTs participated in the study, and each evaluated four EDRs using guided prompts. Open coding and text analysis on 77 EDR evaluations were conducted. Results indicate that 41 EDRs were selected for the evaluation, and PTs’ prior experience in the field, both as a teacher or an observer, influenced their selection process. Features of EDRs considered in decision-making process included types of EDRs, pedagogical and maths skill purpose, ready-made or adaptable, interactive or one-way, and cost. Also, five decision-making criteria used by PTs in EDR selection and use were the primary benefit holder, function of EDRs, opportunity for improving mathematical skill, affordances, and constraints. Findings urged framing a new generations’ perspective on evaluating EDRs. The ways to support PTs for better informed selection and implementation of EDRs were discussed. [ FROM AUTHOR] Copyright of Journal of Education for Teaching is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full . (Copyright applies to all s.)

Elevated resting heart rates are a risk factor for mortality among patients with coronavirus disease 2019 in Wuhan, China.

BACKGROUND: We evaluated the association between higher resting heart rates (RHRs) and adverse events in COVID-19 patients. METHODS: One hundred and thirty-six patients with laboratory-confirmed COVID-19 were admitted. Outcomes of patients with different RHRs were compared. RESULTS: Twenty-nine patients had RHRs of <80 bpm (beat per min), 85 had 80-99 bpm and 22 had ≥100 bpm as tachycardia. Those with higher RHRs had lower pulse oxygen saturation (SpO2) and higher temperatures, and there was a higher proportion of men upon admission (all P < 0.05). Patients with higher RHRs showed higher white blood cell counts and D-dimer, cardiac troponin I (TnI), N-terminal pro-B-type natriuretic peptide and hypersensitive C-reactive protein levels, but lower albumin levels (all P < 0.05) after admission. During follow-up, 26 patients died (mortality rate, 19.1%). The mortality rate was significantly higher among patients with tachycardia than among the moderate and low RHR groups (all P < 0.001). Kaplan-Meier survival curves showed that the risks of death and ventilation use increased for patients with tachycardia (P < 0.001). Elevated RHR as a continuous variable and a mean RHR as tachycardia were independent risk factors for mortality and ventilator use (all P < 0.05) in the multivariable adjusted Cox proportional hazards regression model. CONCLUSIONS: Elevated average RHRs during the first 3 days of hospitalisation were associated with adverse outcomes in COVID-19 patients. Average RHRs as tachycardia can independently predict all-cause mortality.

Etiological and pathophysiological enigmas of severe coronavirus disease 2019, multisystem inflammatory syndrome in children, and Kawasaki disease.

During the coronavirus disease 2019 (COVID-19) pandemic, a novel multisystem inflammatory syndrome in children (MIS-C) has been reported worldwide since the first cases were reported in Europe in April 2020. MIS-C is temporally associated with severe acute respiratory syndrome coronavirus 2 infection and shows Kawasaki disease (KD)-like features. The epidemiological and clinical characteristics in COVID-19, KD, and MIS-C differ, but severe cases of each disease share similar clinical and laboratory findings such as a protracted clinical course, multiorgan involvement, and similar activated biomarkers. These findings suggest that a common control system of the host may act against severe disease insult. To solve the enigmas, we proposed the protein-homeostasis-system hypothesis in that every disease involves etiological substances and the host's immune system controls them by their size and biochemical properties. Also, it is proposed that the etiological agents of KD and MIS-C might be certain strains in the microbiota of human species and etiological substances in severe COVID-19, KD, and MIS-C originate from pathogen-infected cells. Since disease severity depends on the amounts of inflammation-inducing substances and corresponding immune activation in the early stage of the disease, an early proper dose of corticosteroids and/or intravenous immunoglobulin (IVIG) may help reduce morbidity and possibly mortality among patients with these diseases. Corticosteroids are low cost and an analogue of host-origin cortisol among immune modulators. This study's findings will help clinicians treating severe COVID-19, KD, and MIS-C, especially in developing countries, where IVIG and biologics supplies are insufficient.

Joint Predictive Value of cTnI and NT-proBNP on Mortality in Patients with Coronavirus Disease 2019: A Retrospective Research in Wuhan, China.

BACKGROUND AND OBJECTIVES: The pandemic of coronavirus disease 2019 (COVID-19) remains to be the biggest public threat all over the world. Because of the rapid deterioration in some patients, markers that could predict poor clinical outcomes are urgently required. This study was to evaluate the predictive values of cardiac injury parameters, including cardiac troponin I (cTnI) and N-terminal pro-B-type natriuretic peptide (NT-proBNP) levels, on mortality in COVID-19 patients. METHODS: COVID-19 patients in Zhongfaxincheng branch of Tongji Hospital (Wuhan, China) from February 8-28, 2020, were enrolled in this study. We followed up the patients for 30 days after admission. RESULTS: A total of 134 patients were included in the study. Multivariate Cox regression showed that 1) patients with elevated cTnI levels had a higher risk of death (hazard ratio [HR] 7.33, 95% confidence interval [CI] 2.56-21.00) than patients with normal cTnI levels; 2) patients with elevated NT-proBNP levels had a higher risk of death (HR 27.88, 95% CI 3.55-218.78) than patients with normal NT-proBNP levels; 3) patients with both elevated cTnI and NT-proBNP levels had a significantly higher risk of death (HR 53.87, 95% CI 6.31-459.91, P < 0.001) compared to patients without elevated cTnI or NT-proBNP levels; 4) the progressions of cTnI and NT-proBNP levels were also correlated with death (HR 12.70, 95% CI 3.94-40.88, P < 0.001 and HR 51.09, 95% CI 5.82-448.26, P < 0.001). CONCLUSIONS: In COVID-19 patients, cTnI and NT-proBNP levels could be monitored to identify patients at a high risk of death in their later course of disease.

Real-Time Mild and Moderate COVID-19 Human Body Temperature Detection Using Artificial Intelligence Coronavirus Antigens: Efficient Human Cell Coexpression System and Its Application to the Production of Multiple Coronavirus Antigens (Adv. Biology 4/2021)

Summary Day by day, COVID-19 cases are increasing all over the world. Without a proper vaccine to control the disease, the only solution so far is social distancing and identifying the disease at an early stage. In more than 80% of confirmed cases there are only mild symptoms, like fever;therefore, we have to check the body temperature of people in public places like shopping malls, hotels, airports, schools and universities, etc. In this chapter we propose contactless temperature (CT) measurement utilizing thermal (TS), RGB, and 3D sensors. We also propose a fever location camera (FLC) which gives high-quality estimates from up to 2 or 3 meters away. Using cutting-edge technology, the fever location framework (FLF) estimates the internal heat level of individuals in groups of three or four by checking and filtering their face temperatures. If a high temperature is identified, the framework sounds an alarm or cautioning message, which has propelled face recognition technology. The framework, which is based on the investigation of face temperature, guarantees high-quality estimations. Using facial recognition (FR) likewise limits false readings;for example, an individual carrying a hot beverage. Using a devoted programming stage, a signal can be set to inform us of unusual temperatures. It can precisely recognize the facial temperature (FT) of numerous individuals quickly, with an exactness of ≤ 0.3 °C. Temperature recognition range can be set with the ideal location of up to 3 meters in the framework highlighted by a bi-directional double-channel (infrared light + visible light) camera utilizing a heated sensor and low level interference signals. The production of biomolecules that require human-specific lipid environments is extremely useful for basic research and medical applications. In article number 2000154, Seong-Jun Kim, Jae-Sung Woo, Sangsu Bae, and co-workers integrate multiple proteins or virus antigens into defined transcriptional hotspots in the human genome via a homology-independent targeted insertion method using CRISPR nucleases. This system is similar to a production pipeline of biomolecules in a factory controlled by CRISPR.

Coronavirus Antigens: Efficient Human Cell Coexpression System and Its Application to the Production of Multiple Coronavirus Antigens (Adv. Biology 4/2021)

The production of biomolecules that require human-specific lipid environments is extremely useful for basic research and medical applications. In article number 2000154, Seong-Jun Kim, Jae-Sung Woo, Sangsu Bae, and co-workers integrate multiple proteins or virus antigens into defined transcriptional hotspots in the human genome via a homology-independent targeted insertion method using CRISPR nucleases. This system is similar to a production pipeline of biomolecules in a factory controlled by CRISPR.

Inter- and intra-city comparisons of PM2.5 concentration changes under COVID-19 social distancing in seven major cities of South Korea.

The COVID-19 pandemic has prompted governments around the world to impose mitigation strategies of unprecedented scales, typically involving some form of restrictions on social activities and transportation. The South Korean government has been recommending a collection of guidelines now known as social distancing, leading to reduced human activities. This study analyzes changes in the concentrations of fine particulate matter (PM2.5) during the 30-day periods before and since the start of social distancing on 29 February 2020 using measurement data from air quality monitoring stations at various locations of the seven major cities of South Korea, namely, Seoul, Busan, Incheon, Daegu, Daejeon, Gwangju, and Ulsan. All seven cities experienced decreased levels of PM2.5 concentration by up to 25% and smaller fluctuations during the period of social distancing. Inter-city comparisons show that the PM2.5 concentration changes are positively correlated with the city-wide PM2.5 emission fractions for mobile sources and negatively correlated with the city-wide PM2.5 emission fractions for combustion and industrial process sources. In addition, the meteorological influences favorable for transboundary pollutant transport have weakened during the period under COVID-19 social distancing. Intra-city comparisons show that decreases in the intra-city variability of PM2.5 concentration were larger in coastal cities than in inland cities. Comparisons between the inter- and intra-city variabilities in the PM2.5 concentration changes under social distancing highlight the importance of taking into account intra-city variabilities in addition to inter-city variabilities.

Association Between Average Plasma Potassium Levels and 30-day Mortality During Hospitalization in Patients with COVID-19 in Wuhan, China.

Background: Coronavirus disease 2019 (COVID-19) has resulted in more than 610,000 deaths worldwide since December 2019. Given the rapid deterioration of patients' condition before death, markers with efficient prognostic values are urgently required. During the treatment process, notable changes in plasma potassium levels have been observed among severely ill patients. We aimed to evaluate the association between average plasma potassium (Ka +) levels during hospitalization and 30-day mortality in patients with COVID-19. Methods: Consecutive patients with COVID-19 hospitalized in the Zhongfaxincheng branch of Tongji Hospital in Wuhan, China from February 8 to 28, 2020 were enrolled in this study. We followed patients up to 30 days after admission. Results: A total of 136 patients were included in the study. The average age was 62.1±14.6 years and 51.5% of patients were male. The median baseline potassium level was 4.3 (3.9-4.6) mmol/L and Ka + level during hospitalization was 4.4 (4.2-4.7) mmol/L; the median number of times that we measured potassium was 4 (3-5). The 30-day mortality was 19.1%. A J-shaped association was observed between Ka + and 30-day mortality. Multivariate Cox regression showed that compared with the reference group (Ka + 4.0 to <4.5 mmol/L), 30-day mortality was 1.99 (95% confidence interval [CI]=0.54-7.35, P=0.300), 1.14 (95% CI=0.39-3.32, P=0.810), and 4.14 (95% CI=1.29-13.29, P=0.017) times higher in patients with COVID-19 who had Ka + <4.0, 4.5 to <5.0, and ≥5.0 mmol/L, respectively. Conclusion: Patients with COVID-19 who had a Ka + level ≥5.0 mmol/L had a significantly increased 30-day mortality compared with those who had a Ka + level 4.0 to <4.5 mmol/L. Plasma potassium levels should be monitored routinely and maintained within appropriate ranges in patients with COVID-19.

Air Quality Change in Seoul, South Korea under COVID-19 Social Distancing: Focusing on PM2.5.

Seoul, the most populous city in South Korea, has been practicing social distancing to slow down the spread of coronavirus disease 2019 (COVID-19). Fine particulate matter (PM2.5) and other air pollutants measured in Seoul over the two 30 day periods before and after the start of social distancing are analyzed to assess the change in air quality during the period of social distancing. The 30 day mean PM2.5 concentration decreased by 10.4% in 2020, which is contrasted with an average increase of 23.7% over the corresponding periods in the previous 5 years. The PM2.5 concentration decrease was city-wide and more prominent during daytime than at nighttime. The concentrations of carbon monoxide (CO) and nitrogen dioxide (NO2) decreased by 16.9% and 16.4%, respectively. These results show that social distancing, a weaker forcing toward reduced human activity than a strict lockdown, can help lower pollutant emissions. At the same time, synoptic conditions and the decrease in aerosol optical depth over the regions to the west of Seoul support that the change in Seoul's air quality during the COVID-19 social distancing can be interpreted as having been affected by reductions in the long-range transport of air pollutants as well as local emission reductions.

Serum Procalcitonin Levels on Admission Predict Death in Severe and Critical COVID-19 Patients in Wuhan, China

Background: We evaluated whether the serum procalcitonin (PCT) level could predict death in severe and critical coronavirus disease 2019 (COVID-19) patients. Methods: This study included 129 COVID-19 patients. PCT levels on admission, treatment, and death were collected. The outcomes were compared. Results: The optimum cutoff value of the PCT level determined by receiver operator characteristic curve analysis to predict all-cause death was 0.085 ng/mL, with sensitivity of 95.7% and specificity of 72.6%. Overall, 78 patients had a PCT level below 0.085 ng/mL and 51 patients had a PCT level of 0.085 ng/mL or greater. High-PCT-level patients had lower levels of lymphocytes (P= 0.001) and albumin (P = 0.002) and higher levels of creatinine (P = 0.024), D-dimer (P = 0.002), and white blood cells, neutrocytes (P < 0.001), high-sensitivity C-reactive protein (P < 0.001), interleukin-6 (P < 0.001), interleukin-8 (P = 0.001), interleukin-10 (P = 0.001), tumor necrosis factor (P < 0.001), erythrocyte sedimentation rate (P = 0.001), and ferritin (P = 0.001). During the 30-day observation period, 23 patients died. Mortality was significantly higher in high-PCT-level patients than in patients with low PCT levels (43.1% vs. 1.3%;P < 0.001). The risks of death (P < 0.0001) and ventilator use (P < 0.0001) were increased in patients with PCT levels of 0.085 ng/mL or greater. Conclusions: A PCT level of 0.085 ng/mL or greater on admission could effectively predict death and ventilator use in severe and critical COVID-19 patients.

Clinical Spectrum of Sickle Cell Disease and COVID-19: Laboratory and Clinical Factors Associated with Morbidity and Mortality (preprint)

Background: The COVID-19 pandemic has generated concern as a potential remarkably severe threat to the sickle cell disease (SCD) population. We aim to identify predictors of outcomes and survival in a large US-based SCD and COVID-19 cohort to inform best approaches to prevention and care.Methods: Clinical data were collected at baseline and during the clinical course using a standardized form in SCD patients diagnosed with COVID-19 at 5 academic centers in four COVID-19 epicenters. Patients were followed post-hospital discharge for up to 3 months.Results: Of 66 consecutive SCD patients with COVID-19, 75% required hospitalization, with a median length of stay of 6 days, and 7 died (10.6%). Patients with preexisting kidney disease were more likely to be hospitalized, while age, sex and genotype had no effect. The most common presenting symptom was vaso-occlusive pain. Chest X-ray was abnormal (acute chest syndrome) at presentation in 30 of 50 (61%) hospitalized and all deceased patients. Older age (median of 53 versus 32 years) and histories of pulmonary hypertension, congestive heart failure and/or stroke were more prevalent in deceased patients, as were high creatinine, lactate dehydrogenase, C-reactive protein, and D-dimers. The use of anti-coagulation, but not hydroxychloroquine or transfusion, during inpatient hospitalization was associated with decreased mortality (p<0.05). All deaths occurred in individuals not taking hydroxyurea or other SCD modifying therapy.Conclusions: Patients with SCD and COVID-19 infection demonstrated a broad range of disease severity, from mild to very severe. COVID-19 in SCD individuals with pre-existing cardiopulmonary, renal disease and/or stroke presenting with pain and high creatinine should be considered at risk of death, irrespective of genotype or gender. Inpatient use of anticoagulation should be considered for all SCD patients with COVID-19. Though older individuals with vasculopathic comorbidities and high D-dimers were more likely to die, the median age of death was decades lower than the non-SCD population.Funding Statement: None.Declaration of Interests: NoneEthics Approval Statement: Each respective Institutional Review Board approved data collection as minimal-risk research and waived the requirement for informed consent.

Guidelines for Surgery of Confirmed or Suspected COVID-19 Patients

Coronavirus disease 2019 (COVID-19) has spread widely across the world since January 2020. There are many challenges when caring for patients with COVID-19, one of which is infection prevention and control. In particular, in cases where surgery must absolutely be performed, special infection control may be required in order to perform surgery without spreading infection within the hospital. We aim to present potentially useful recommendations for non-deferrable surgery for COVID-19 patients based on in vivo and in vitro research and clinical experiences from many countries.

Probabilistic Neighbourhood Component Analysis: Sample Efficient Uncertainty Estimation in Deep Learning (preprint)

While Deep Neural Networks (DNNs) achieve state-of-the-art accuracy in various applications, they often fall short in accurately estimating their predictive uncertainty and, in turn, fail to recognize when these predictions may be wrong. Several uncertainty-aware models, such as Bayesian Neural Network (BNNs) and Deep Ensembles have been proposed in the literature for quantifying predictive uncertainty. However, research in this area has been largely confined to the big data regime. In this work, we show that the uncertainty estimation capability of state-of-the-art BNNs and Deep Ensemble models degrades significantly when the amount of training data is small. To address the issue of accurate uncertainty estimation in the small-data regime, we propose a probabilistic generalization of the popular sample-efficient non-parametric kNN approach. Our approach enables deep kNN classifier to accurately quantify underlying uncertainties in its prediction. We demonstrate the usefulness of the proposed approach by achieving superior uncertainty quantification as compared to state-of-the-art on a real-world application of COVID-19 diagnosis from chest X-Rays. Our code is available at https://github.com/ankurmallick/sample-efficient-uq

Immunopathogenesis of COVID-19 and early immunomodulators.

The novel coronavirus disease 2019 (COVID-19) is spreading globally. Although its etiologic agent is discovered as severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), there are many unsolved issues in COVID-19 and other infectious diseases. The causes of different clinical phenotypes and incubation periods among individuals, species specificity, and cytokine storm with lymphopenia as well as the mechanism of damage to organ cells are unknown. It has been suggested that in viral pneumonia, virus itself is not a direct cause of acute lung injury; rather, aberrant immune reactions of the host to the insults from viral infection are responsible. According to its epidemiological and clinical characteristics, SARS-CoV-2 may be a virus with low virulence in nature that has adapted to the human species. Current immunological concepts have limited ability to explain such unsolved issues, and a presumed immunopathogenesis of COVID-19 is presented under the proteinhomeostasis-system hypothesis. Every disease, including COVID-19, has etiological substances controlled by the host immune system according to size and biochemical properties. Patients with severe pneumonia caused by SARS-CoV-2 show more severe hypercytokinemia with corresponding lymphocytopenia than patients with mild pneumonia; thus, early immunomodulator treatment, including corticosteroids, has been considered. However, current guidelines recommend their use only for patients with advanced pneumonia or acute respiratory distress syndrome. Since the immunopathogenesis of pneumonia may be the same for all patients regardless of age or severity and the critical immune-mediated lung injury may begin in the early stage of the disease, early immunomodulator treatment, including corticosteroids and intravenous immunoglobulin, can help reduce morbidity and possibly mortality rates of older patients with underlying conditions.