Inosine: a broad-spectrum anti-inflammatory against SARS-CoV-2 infection-induced acute lung injury via suppressing TBK1 phosphorylation.

SARS-CoV-2-induced cytokine storms constitute the primary cause of COVID-19 progression, severity, criticality, and death. Glucocorticoid and anti-cytokine therapies have been frequently administered to treat COVID-19 but have had limited clinical efficacy in severe and critical cases. Nevertheless, the weaknesses of these treatment modalities have prompted the development of anti-inflammatory therapy against this infection. We found that the broad-spectrum anti-inflammatory agent inosine downregulated proinflammatory IL-6, upregulated anti-inflammatory IL-10, and ameliorated acute inflammatory lung injury caused by multiple infectious agents. Inosine significantly improved survival in mice infected with SARS-CoV-2. It indirectly impeded TANK-binding kinase 1 (TBK1) phosphorylation by binding stimulator of interferon genes (STING) and glycogen synthase kinase-3ß (GSK3ß), inhibited the activation and nuclear translocation of the downstream transcription factors IRF3 and NF-κB, and downregulated IL-6 in the sera and lung tissues of mice infected with lipopolysaccharide (LPS), H1N1, or SARS-CoV-2. Thus, inosine administration is feasible for clinical anti-inflammatory therapy against severe and critical COVID-19. Moreover, targeting TBK1 is a promising strategy for inhibiting cytokine storms and mitigating acute inflammatory lung injury induced by SARS-CoV-2 and other infectious agents.

GSK3326595 is a promising drug to prevent SARS-CoV-2 Omicron and other variants infection by inhibiting ACE2-R671 di-methylation.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) caused COVID-19 epidemic is worsening. Binding of the Spike1 protein of SARS-CoV-2 with the angiotensin-converting enzyme 2 (ACE2) receptor mediates entry of the virus into host cells. Many reports show that protein arginine methylation by protein arginine methyltransferases (PRMTs) is important for the functions of these proteins, but it remains unclear whether ACE2 is methylated by PRMTs. Here, we show that PRMT5 catalyses ACE2 symmetric dimethylation at residue R671 (meR671-ACE2). We indicate that PRMT5-mediated meR671-ACE2 promotes SARS-CoV-2 receptor-binding domain (RBD) binding with ACE2 probably by enhancing ACE2 N-glycosylation modification. We also reveal that the PRMT5-specific inhibitor GSK3326595 is able to dramatically reduce ACE2 binding with RBD. Moreover, we discovered that meR671-ACE2 plays an important role in ACE2 binding with Spike1 of the SARS-CoV-2 Omicron, Delta, and Beta variants; and we found that GSK3326595 strongly attenuates ACE2 interaction with Spike1 of the SARS-CoV-2 Omicron, Delta, and Beta variants. Finally, SARS-CoV-2 pseudovirus infection assays uncovered that PRMT5-mediated meR671-ACE2 is essential for SARS-CoV-2 infection in human cells, and pseudovirus infection experiments confirmed that GSK3326595 can strongly suppress SARS-CoV-2 infection of host cells. Our findings suggest that as a clinical phase II drug for several kinds of cancers, GSK3326595 is a promising candidate to decrease SARS-CoV-2 infection by inhibiting ACE2 methylation and ACE2-Spike1 interaction.

Prevalence of post-traumatic stress disorder symptoms among patients with mental disorder during the COVID-19 pandemic.

BACKGROUND: The outbreak of the COVID-19 pandemic has caused extensive public health concerns, posing significant challenges to healthcare services. One particular area of concern is the mental health of patients with mental disorder, who are often a neglected group. The aim of this study was to investigate the prevalence of, and associated factors for symptoms of post-traumatic stress disorder (PTSD) among patients with mental disorder in China during the COVID-19 pandemic. METHODS: Self-reported questionnaires were distributed to patients in four psychiatric hospitals in Beijing, China, between April 28th and May 30th, 2020. Information regarding sociodemographic characteristics, COVID-19 related factors, support, psychosomatic factors, and PTSD symptoms were collected using a series of scales, such as the Impact of Event Scale-Revised, the 7-item Generalized Anxiety Disorder Scale, the 9-item Patient Health Questionnaire depression scale, and so on. Multivariate regression was used to identify factors related to PTSD symptoms. RESULTS: A total of 1,055 patients with mental disorder were included in the final sample. The prevalence of PTSD symptoms was 41.3%. Hierarchical linear regression demonstrated that fear of the pandemic and anxiety were shared associated factors for both symptoms of PTSD and their subscales. Additionally, age was an associated factor for the total PTSD (ß = 0.12, p < 0.01), intrusion (ß = 0.18, p < 0.001), and avoidance (ß = 0.1, p < 0.05) symptoms; depression was an associated factor for the total PTSD (ß = 0.13, p < 0.001), intrusion (ß = 0.11, p < 0.01), and hyperarousal (ß = 0.19, p < 0.001) symptoms. CONCLUSIONS: The prevalence of PTSD symptoms was high among patients with mental disorder during the COVID-19 pandemic in China. This study found that age, fear of the pandemic, anxiety and depression are significant associated factors of PTSD symptoms in patients with mental disorder during the pandemic. We call for higher awareness and introduction of PTSD interventions to relieve the psychological stress in these patients.

Baicalein self-microemulsion based on drug-phospholipid complex for the alleviation of cytokine storm.

Cytokine storm is a phenomenon whereby the overreaction of the human immune system leads to the release of inflammatory cytokines, which can lead to multiple organ dysfunction syndrome. At present, the existing drugs for the treatment of cytokine storm have limited efficacy and severe adverse effects. Here, we report a lymphatic targeting self-microemulsifying drug delivery system containing baicalein to effectively inhibit cytokine storm. Baicalein self-microemulsion with phospholipid complex as an intermediate carrier (BAPC-SME) prepared in this study could be spontaneously emulsified to form 12-nm oil-in-water nanoemulsion after administration. And then BAPC-SME underwent uptake by enterocyte through endocytosis mediated by lipid valve and clathrin, and had obvious characteristics of mesenteric lymph node targeting distribution. Oral administration of BAPC-SME could significantly inhibit the increase in plasma levels of 14 cytokines: TNF-α, IL-6, IFN-γ, MCP-1, IL-17A, IL-27, IL-1α, GM-CSF, MIG, IFN-ß, IL-12, MIP-3α, IL-23, and RANTES in mice experiencing systemic cytokine storm. BAPC-SME could also significantly improve the pathological injury and inflammatory cell infiltration of lung tissue in mice experiencing local cytokine storm. This study does not only provide a new lymphatic targeted drug delivery strategy for the treatment of cytokine storm but also has great practical significance for the clinical development of baicalein self-microemulsion therapies for cytokine storm.

Post-pandemic healthcare for COVID-19 vaccine: Tissue-aware diagnosis of cervical lymphadenopathy via multi-modal ultrasound semantic segmentation.

With the widespread deployment of COVID-19 vaccines all around the world, billions of people have benefited from the vaccination and thereby avoiding infection. However, huge amount of clinical cases revealed diverse side effects of COVID-19 vaccines, among which cervical lymphadenopathy is one of the most frequent local reactions. Therefore, rapid detection of cervical lymph node (LN) is essential in terms of vaccine recipients' healthcare and avoidance of misdiagnosis in the post-pandemic era. This paper focuses on a novel deep learning-based framework for the rapid diagnosis of cervical lymphadenopathy towards COVID-19 vaccine recipients. Existing deep learning-based computer-aided diagnosis (CAD) methods for cervical LN enlargement mostly only depend on single modal images, e.g., grayscale ultrasound (US), color Doppler ultrasound, and CT, while failing to effectively integrate information from the multi-source medical images. Meanwhile, both the surrounding tissue objects of the cervical LNs and different regions inside the cervical LNs may imply valuable diagnostic knowledge which is pending for mining. In this paper, we propose an Tissue-Aware Cervical Lymph Node Diagnosis method (TACLND) via multi-modal ultrasound semantic segmentation. The method effectively integrates grayscale and color Doppler US images and realizes a pixel-level localization of different tissue objects, i.e., lymph, muscle, and blood vessels. With inter-tissue and intra-tissue attention mechanisms applied, our proposed method can enhance the implicit tissue-level diagnostic knowledge in both spatial and channel dimension, and realize diagnosis of cervical LN with normal, benign or malignant state. Extensive experiments conducted on our collected cervical LN US dataset demonstrate the effectiveness of our methods on both tissue detection and cervical lymphadenopathy diagnosis. Therefore, our proposed framework can guarantee efficient diagnosis for the vaccine recipients' cervical LN, and assist doctors to discriminate between COVID-related reactive lymphadenopathy and metastatic lymphadenopathy.

Assessment of Changes in Visits and Antibiotic Prescribing During the Agency for Healthcare Research and Quality Safety Program for Improving Antibiotic Use and the COVID-19 Pandemic.

Importance: The Agency for Healthcare Research and Quality (AHRQ) Safety Program for Improving Antibiotic Use aimed to improve antibiotic prescribing in ambulatory care practices by engaging clinicians and staff to incorporate antibiotic stewardship into practice culture, communication, and decision-making. Little is known about implementation of antibiotic stewardship in ambulatory care practices. Objective: To examine changes in visits and antibiotic prescribing during the AHRQ Safety Program. Design, Setting, and Participants: This cohort study evaluated a quality improvement intervention in ambulatory care throughout the US in 389 ambulatory care practices from December 1, 2019, to November 30, 2020. Exposures: The AHRQ Safety Program used webinars, audio presentations, educational tools, and office hours to engage stewardship leaders and clinical staff to address attitudes and cultures that challenge judicious antibiotic prescribing and incorporate best practices for the management of common infections. Main Outcomes and Measures: The primary outcome of the Safety Program was antibiotic prescriptions per 100 acute respiratory infection (ARI) visits. Data on total visits and ARI visits were also collected. The number of visits and prescribing rates from baseline (September 1, 2019) to completion of the program (November 30, 2020) were compared. Results: Of 467 practices enrolled, 389 (83%) completed the Safety Program; of these, 292 (75%) submitted complete data with 6 590 485 visits to 5483 clinicians. Participants included 82 (28%) primary care practices, 103 (35%) urgent care practices, 34 (12%) federally supported practices, 39 (13%) pediatric urgent care practices, 21 (7%) pediatric-only practices, and 14 (5%) other practice types. Visits per practice per month decreased from a mean of 1624 (95% CI, 1317-1931) at baseline to a nadir of 906 (95% CI, 702-1111) early in the COVID-19 pandemic (April 2020), and were 1797 (95% CI, 1510-2084) at the end of the program. Total antibiotic prescribing decreased from 18.2% of visits at baseline to 9.5% at completion of the program (-8.7%; 95% CI, -9.9% to -7.6%). Acute respiratory infection visits per practice per month decreased from baseline (n = 321) to a nadir of 76 early in the pandemic (May 2020) and gradually increased through completion of the program (n = 239). Antibiotic prescribing for ARIs decreased from 39.2% at baseline to 24.7% at completion of the program (-14.5%; 95% CI, -16.8% to -12.2%). Conclusions and Relevance: In this study of US ambulatory practices that participated in the AHRQ Safety Program, significant reductions in the rates of overall and ARI-related antibiotic prescribing were noted, despite normalization of clinic visits by completion of the program. The forthcoming AHRQ Safety Program content may have utility in ambulatory practices across the US.

Assessment of Awake Prone Positioning in Hospitalized Adults With COVID-19: A Nonrandomized Controlled Trial.

Importance: Awake prone positioning may improve hypoxemia among patients with COVID-19, but whether it is associated with improved clinical outcomes remains unknown. Objective: To determine whether the recommendation of awake prone positioning is associated with improved outcomes among patients with COVID-19-related hypoxemia who have not received mechanical ventilation. Design, Setting, and Participants: This pragmatic nonrandomized controlled trial was conducted at 2 academic medical centers (Vanderbilt University Medical Center and NorthShore University HealthSystem) during the COVID-19 pandemic. A total of 501 adult patients with COVID-19-associated hypoxemia who had not received mechanical ventilation were enrolled from May 13 to December 11, 2020. Interventions: Patients were assigned 1:1 to receive either the practitioner-recommended awake prone positioning intervention (intervention group) or usual care (usual care group). Main Outcomes and Measures: Primary outcome analyses were performed using a bayesian proportional odds model with covariate adjustment for clinical severity ranking based on the World Health Organization ordinal outcome scale, which was modified to highlight the worst level of hypoxemia on study day 5. Results: A total of 501 patients (mean [SD] age, 61.0 [15.3] years; 284 [56.7%] were male; and most [417 (83.2%)] were self-reported non-Hispanic or non-Latinx) were included. Baseline severity was comparable between the intervention vs usual care groups, with 170 patients (65.9%) vs 162 patients (66.7%) receiving oxygen via standard low-flow nasal cannula, 71 patients (27.5%) vs 62 patients (25.5%) receiving oxygen via high-flow nasal cannula, and 16 patients (6.2%) vs 19 patients (7.8%) receiving noninvasive positive-pressure ventilation. Nursing observations estimated that patients in the intervention group spent a median of 4.2 hours (IQR, 1.8-6.7 hours) in the prone position per day compared with 0 hours (IQR, 0-0.7 hours) per day in the usual care group. On study day 5, the bayesian posterior probability of the intervention group having worse outcomes than the usual care group on the modified World Health Organization ordinal outcome scale was 0.998 (posterior median adjusted odds ratio [aOR], 1.63; 95% credibility interval [CrI], 1.16-2.31). However, on study days 14 and 28, the posterior probabilities of harm were 0.874 (aOR, 1.29; 95% CrI, 0.84-1.99) and 0.673 (aOR, 1.12; 95% CrI, 0.67-1.86), respectively. Exploratory outcomes (progression to mechanical ventilation, length of stay, and 28-day mortality) did not differ between groups. Conclusions and Relevance: In this nonrandomized controlled trial, prone positioning offered no observed clinical benefit among patients with COVID-19-associated hypoxemia who had not received mechanical ventilation. Moreover, there was substantial evidence of worsened clinical outcomes at study day 5 among patients recommended to receive the awake prone positioning intervention, suggesting potential harm. Trial Registration: ClinicalTrials.gov Identifier: NCT04359797.

PRCTC: a machine learning model for prediction of response to corticosteroid therapy in COVID-19 patients.

Corticosteroid has been proved to be one of the few effective treatments for COVID-19 patients. However, not all the patients were suitable for corticosteroid therapy. In this study, we aimed to propose a machine learning model to forecast the response to corticosteroid therapy in COVID-19 patients. We retrospectively collected the clinical data about 666 COVID-19 patients receiving corticosteroid therapy between January 27, 2020, and March 30, 2020, from two hospitals in China. The response to corticosteroid therapy was evaluated by hospitalization time, oxygen supply duration, and the outcomes of patients. Least Absolute Shrinkage and Selection Operator (LASSO) was applied for feature selection. Five prediction models were applied in the training cohort and assessed in an internal and an external validation dataset, respectively. Finally, two (C reactive protein, lymphocyte percent) of 36 candidate immune/inflammatory features were finally used for model development. All five models displayed promising predictive performance. Notably, the ensemble model, PRCTC (prediction of response to corticosteroid therapy in COVID-19 patients), derived from three prediction models including Gradient Boosted Decision Tree (GBDT), Neural Network (NN), and logistic regression (LR), achieved the best performance with an area under the curve (AUC) of 0.810 (95% confidence interval [CI] 0.760-0.861) in internal validation cohort and 0.845 (95% CI 0.779-0.911) in external validation cohort to predict patients' response to corticosteroid therapy. In conclusion, PRCTC proposed with universality and scalability is hopeful to provide tangible and prompt clinical decision support in management of COVID-19 patients and potentially extends to other medication predictions.

Impact of COVID-19 on presentation, staging, and treatment of head and neck mucosal squamous cell carcinoma.

OBJECTIVES: During the COVID-19 pandemic, maintenance of safe and timely oncologic care has been challenging. The goal of this study is to compare presenting symptoms, staging, and treatment of head and neck mucosal squamous cell carcinoma during the pandemic with an analogous timeframe one year prior. MATERIALS AND METHODS: Retrospective cohort study at a single tertiary academic center of new adult patients evaluated in a head and neck surgical oncology clinic from March -July 2019 (pre-pandemic control) and March - July 2020 (COVID-19 pandemic). RESULTS: During the pandemic, the proportion of patients with newly diagnosed malignancies increased by 5%, while the overall number of new patients decreased (n = 575) compared to the control year (n = 776). For patients with mucosal squamous cell carcinoma (SCC), median time from referral to initial clinic visit decreased from 11 days (2019) to 8 days (2020) (p = 0.0031). There was no significant difference in total number (p = 0.914) or duration (p = 0.872) of symptoms. During the pandemic, patients were more likely to present with regional nodal metastases (adjusted odds ratio (OR) 2.846, 95% CI 1.072-3.219, p = 0.028) and more advanced clinical nodal (N) staging (p = 0.011). No significant difference was seen for clinical tumor (T) (p = 0.502) or metastasis (M) staging (p = 0.278). No significant difference in pathologic T (p = 0.665), or N staging (p = 0.907) was found between the two periods. CONCLUSION: Head and neck mucosal SCC patients presented with more advanced clinical nodal disease during the early months of the COVID-19 pandemic despite no change in presenting symptoms.

Clinical features and survival analysis of 97 coronavirus disease 2019 (COVID-19) patients.

BACKGROUND: We aim to investigate the clinical characteristics and survival rate of coronavirus disease 2019 (COVID-19) patients. METHODS: Ninety-seven COVID-19 patients were enrolled. The laboratory results, lung imaging and medical treatment were compared. Patients were followed up after 1 year, and the Kaplan-Meier test was used for survival analysis. RESULTS: Compared with the non-severe group, the age of the severe group was older, and the proportion of concomitant diseases were higher. As fever was the primary clinical manifestation, dyspnea and anorexia were more common in severe patients. Lung imaging manifestations and laboratory indicators were worse in the severe group. Accordingly, the treatment of glucocorticoid, antibiotics, and advanced life support were in high proportion. Of the 97 patients with COVID-19, 4 severe patients died within one month during the 1-year follow-up, with the median survival time of 47.0 weeks (95% CI: 45.1-48.9). CONCLUSIONS: Severe cases of COVID-19 are characterized by advanced age, more concomitant diseases and complications, which lead to a decreased short-term survival rate. However, there were no deaths after one month, which implied a good prognosis if the risk period were passed smoothly.

Joint segmentation and detection of COVID-19 via a sequential region generation network.

The fast pandemics of coronavirus disease (COVID-19) has led to a devastating influence on global public health. In order to treat the disease, medical imaging emerges as a useful tool for diagnosis. However, the computed tomography (CT) diagnosis of COVID-19 requires experts' extensive clinical experience. Therefore, it is essential to achieve rapid and accurate segmentation and detection of COVID-19. This paper proposes a simple yet efficient and general-purpose network, called Sequential Region Generation Network (SRGNet), to jointly detect and segment the lesion areas of COVID-19. SRGNet can make full use of the supervised segmentation information and then outputs multi-scale segmentation predictions. Through this, high-quality lesion-areas suggestions can be generated on the predicted segmentation maps, reducing the diagnosis cost. Simultaneously, the detection results conversely refine the segmentation map by a post-processing procedure, which significantly improves the segmentation accuracy. The superiorities of our SRGNet over the state-of-the-art methods are validated through extensive experiments on the built COVID-19 database.

Early identification of patients with severe COVID-19 at increased risk of in-hospital death: a multicenter case-control study in Wuhan.

BACKGROUND: Most evidence regarding the risk factors for early in-hospital mortality in patients with severe COVID-19 focused on laboratory data at the time of hospital admission without adequate adjustment for confounding variables. A multicenter, age-matched, case-control study was therefore designed to explore the dynamic changes in laboratory parameters during the first 10 days after admission and identify early risk indicators for in-hospital mortality in this patient cohort. METHODS: Demographics and clinical data were extracted from the medical records of 93 pairs of patients who had been admitted to hospital with severe COVID-19. These patients had either been discharged or were deceased by March 3, 2020. Data from days 1, 4, 7, and 10 of hospital admission were compared between survivors and non-survivors. Univariate and multivariate conditional logistic regression analyses were employed to identify early risk indicators of in-hospital death in this cohort. RESULTS: On admission, in-hospital mortality was associated with five risk indicators (ORs in descending order): aspartate aminotransferase (AST, >32 U/L) 43.20 (95% CI: 2.63, 710.04); C-reactive protein (CRP) greater than 100 mg/L 13.61 (1.78, 103.941); lymphocyte count lower than 0.6×109/L 9.95 (1.30, 76.42); oxygen index (OI) less than 200 8.23 (1.04, 65.15); and D-dimer over 1 mg/L 8.16 (1.23, 54.34). Sharp increases in D-dimer at day 4, accompanied by decreasing lymphocyte counts, deteriorating OI, and persistent remarkably high CRP concentration were observed among non-survivors during the early stages of hospital admission. CONCLUSIONS: The potential risk factors of high D-dimer, CRP, AST, low lymphocyte count and OI could help clinicians identify patients at high risk of death early in the hospital admission. This might assist with rationalization of health care resources.

Development and validation of an online model to predict critical COVID-19 with immune-inflammatory parameters.

BACKGROUND: Immune and inflammatory dysfunction was reported to underpin critical COVID-19(coronavirus disease 2019). We aim to develop a machine learning model that enables accurate prediction of critical COVID-19 using immune-inflammatory features at admission. METHODS: We retrospectively collected 2076 consecutive COVID-19 patients with definite outcomes (discharge or death) between January 27, 2020 and March 30, 2020 from two hospitals in China. Critical illness was defined as admission to intensive care unit, receiving invasive ventilation, or death. Least Absolute Shrinkage and Selection Operator (LASSO) was applied for feature selection. Five machine learning algorithms, including Logistic Regression (LR), Support Vector Machine (SVM), Gradient Boosted Decision Tree (GBDT), K-Nearest Neighbor (KNN), and Neural Network (NN) were built in a training dataset, and assessed in an internal validation dataset and an external validation dataset. RESULTS: Six features (procalcitonin, [T + B + NK cell] count, interleukin 6, C reactive protein, interleukin 2 receptor, T-helper lymphocyte/T-suppressor lymphocyte) were finally used for model development. Five models displayed varying but all promising predictive performance. Notably, the ensemble model, SPMCIIP (severity prediction model for COVID-19 by immune-inflammatory parameters), derived from three contributive algorithms (SVM, GBDT, and NN) achieved the best performance with an area under the curve (AUC) of 0.991 (95% confidence interval [CI] 0.979-1.000) in internal validation cohort and 0.999 (95% CI 0.998-1.000) in external validation cohort to identify patients with critical COVID-19. SPMCIIP could accurately and expeditiously predict the occurrence of critical COVID-19 approximately 20 days in advance. CONCLUSIONS: The developed online prediction model SPMCIIP is hopeful to facilitate intensive monitoring and early intervention of high risk of critical illness in COVID-19 patients. TRIAL REGISTRATION: This study was retrospectively registered in the Chinese Clinical Trial Registry ( ChiCTR2000032161 ). vv.

Immunological alternation in COVID-19 patients with cancer and its implications on mortality.

Patients with malignancy were reportedly more susceptible and vulnerable to Coronavirus Disease 2019 (COVID-19), and witnessed a greater mortality risk in COVID-19 infection than noncancerous patients. But the role of immune dysregulation of malignant patients on poor prognosis of COVID-19 has remained insufficiently investigated. Here we conducted a retrospective cohort study that included 2,052 patients hospitalized with COVID-19 (Cancer, n = 93; Non-cancer, n = 1,959), and compared the immunological characteristics of both cohorts. We used stratification analysis, multivariate regressions, and propensity-score matching to evaluate the effect of immunological indices. In result, COVID-19 patients with cancer had ongoing and significantly elevated inflammatory factors and cytokines (high-sensitivity C-reactive protein, procalcitonin, interleukin (IL)-2 receptor, IL-6, IL-8), as well as decreased immune cells (CD8 + T cells, CD4 + T cells, B cells, NK cells, Th and Ts cells) than those without cancer. The mortality rate was significantly higher in cancer cohort (24.7%) than non-cancer cohort (10.8%). By stratification analysis, COVID-19 patients with immune dysregulation had poorer prognosis than those with the relatively normal immune system both in cancer and non-cancer cohort. By logistic regression, Cox regression, and propensity-score matching, we found that prior to adjustment for immunological indices, cancer history was associated with an increased mortality risk of COVID-19 (p < .05); after adjustment for immunological indices, cancer history was no longer an independent risk factor for poor prognosis of COVID-19 (p > .30). In conclusion, COVID-19 patients with cancer had more severely dysregulated immune responses than noncancerous patients, which might account for their poorer prognosis. Clinical Trial: This study has been registered on the Chinese Clinical Trial Registry (No. ChiCTR2000032161).

Accurately Differentiating Between Patients With COVID-19, Patients With Other Viral Infections, and Healthy Individuals: Multimodal Late Fusion Learning Approach.

BACKGROUND: Effectively identifying patients with COVID-19 using nonpolymerase chain reaction biomedical data is critical for achieving optimal clinical outcomes. Currently, there is a lack of comprehensive understanding in various biomedical features and appropriate analytical approaches for enabling the early detection and effective diagnosis of patients with COVID-19. OBJECTIVE: We aimed to combine low-dimensional clinical and lab testing data, as well as high-dimensional computed tomography (CT) imaging data, to accurately differentiate between healthy individuals, patients with COVID-19, and patients with non-COVID viral pneumonia, especially at the early stage of infection. METHODS: In this study, we recruited 214 patients with nonsevere COVID-19, 148 patients with severe COVID-19, 198 noninfected healthy participants, and 129 patients with non-COVID viral pneumonia. The participants' clinical information (ie, 23 features), lab testing results (ie, 10 features), and CT scans upon admission were acquired and used as 3 input feature modalities. To enable the late fusion of multimodal features, we constructed a deep learning model to extract a 10-feature high-level representation of CT scans. We then developed 3 machine learning models (ie, k-nearest neighbor, random forest, and support vector machine models) based on the combined 43 features from all 3 modalities to differentiate between the following 4 classes: nonsevere, severe, healthy, and viral pneumonia. RESULTS: Multimodal features provided substantial performance gain from the use of any single feature modality. All 3 machine learning models had high overall prediction accuracy (95.4%-97.7%) and high class-specific prediction accuracy (90.6%-99.9%). CONCLUSIONS: Compared to the existing binary classification benchmarks that are often focused on single-feature modality, this study's hybrid deep learning-machine learning framework provided a novel and effective breakthrough for clinical applications. Our findings, which come from a relatively large sample size, and analytical workflow will supplement and assist with clinical decision support for current COVID-19 diagnostic methods and other clinical applications with high-dimensional multimodal biomedical features.

Hypergraph learning for identification of COVID-19 with CT imaging.

The coronavirus disease, named COVID-19, has become the largest global public health crisis since it started in early 2020. CT imaging has been used as a complementary tool to assist early screening, especially for the rapid identification of COVID-19 cases from community acquired pneumonia (CAP) cases. The main challenge in early screening is how to model the confusing cases in the COVID-19 and CAP groups, with very similar clinical manifestations and imaging features. To tackle this challenge, we propose an Uncertainty Vertex-weighted Hypergraph Learning (UVHL) method to identify COVID-19 from CAP using CT images. In particular, multiple types of features (including regional features and radiomics features) are first extracted from CT image for each case. Then, the relationship among different cases is formulated by a hypergraph structure, with each case represented as a vertex in the hypergraph. The uncertainty of each vertex is further computed with an uncertainty score measurement and used as a weight in the hypergraph. Finally, a learning process of the vertex-weighted hypergraph is used to predict whether a new testing case belongs to COVID-19 or not. Experiments on a large multi-center pneumonia dataset, consisting of 2148 COVID-19 cases and 1182 CAP cases from five hospitals, are conducted to evaluate the prediction accuracy of the proposed method. Results demonstrate the effectiveness and robustness of our proposed method on the identification of COVID-19 in comparison to state-of-the-art methods.

Machine learning based early warning system enables accurate mortality risk prediction for COVID-19.

Soaring cases of coronavirus disease (COVID-19) are pummeling the global health system. Overwhelmed health facilities have endeavored to mitigate the pandemic, but mortality of COVID-19 continues to increase. Here, we present a mortality risk prediction model for COVID-19 (MRPMC) that uses patients' clinical data on admission to stratify patients by mortality risk, which enables prediction of physiological deterioration and death up to 20 days in advance. This ensemble model is built using four machine learning methods including Logistic Regression, Support Vector Machine, Gradient Boosted Decision Tree, and Neural Network. We validate MRPMC in an internal validation cohort and two external validation cohorts, where it achieves an AUC of 0.9621 (95% CI: 0.9464-0.9778), 0.9760 (0.9613-0.9906), and 0.9246 (0.8763-0.9729), respectively. This model enables expeditious and accurate mortality risk stratification of patients with COVID-19, and potentially facilitates more responsive health systems that are conducive to high risk COVID-19 patients.