SERVAL: Synergy Learning between Vertical Models and LLMs towards Oracle-Level Zero-shot Medical Prediction (preprint)

Recent development of large language models (LLMs) has exhibited impressive zero-shot proficiency on generic and common sense questions. However, LLMs' application on domain-specific vertical questions still lags behind, primarily due to the humiliation problems and deficiencies in vertical knowledge. Furthermore, the vertical data annotation process often requires labor-intensive expert involvement, thereby presenting an additional challenge in enhancing the model's vertical capabilities. In this paper, we propose SERVAL, a synergy learning pipeline designed for unsupervised development of vertical capabilities in both LLMs and small models by mutual enhancement. Specifically, SERVAL utilizes the LLM's zero-shot outputs as annotations, leveraging its confidence to teach a robust vertical model from scratch. Reversely, the trained vertical model guides the LLM fine-tuning to enhance its zero-shot capability, progressively improving both models through an iterative process. In medical domain, known for complex vertical knowledge and costly annotations, comprehensive experiments show that, without access to any gold labels, SERVAL with the synergy learning of OpenAI GPT-3.5 and a simple model attains fully-supervised competitive performance across ten widely used medical datasets. These datasets represent vertically specialized medical diagnostic scenarios (e.g., diabetes, heart diseases, COVID-19), highlighting the potential of SERVAL in refining the vertical capabilities of LLMs and training vertical models from scratch, all achieved without the need for annotations.

A case of multiple thrombosis and septic shock in a critically ill patient with Omicron infection

The emergence of the acute respiratory syndrome coronavirus 2 variant named Omicron has become a global concern. A 74-year-old unvaccinated patient was critically ill infected with the Omicron variant characterised by septic shock, large-scale cerebral embolism, deep vein thrombosis, and multiple organ dysfunction with respiratory failure, acute renal failure, coagulation dysfunction. The clinical symptoms were successfully controlled by active rescue treatment such as anti-infection, anti-shock, implantation of a vena cava filter as well as multi-organ function support. Although there are many complications in critically ill patients with Omicron variant infections, especially coagulation disorders and thrombosis, they can be resolved with a combination of Chinese and Western medicine positive rescue.

Estimating daily ground-level NO2 concentrations over China based on TROPOMI observations and machine learning approach

Nitrogen dioxide (NO2) is an important target for monitoring atmospheric quality. Deriving ground-level NO2 concentrations with much finer resolution, it requires high-resolution satellite tropospheric NO2 column as input and a reliable estimation algorithm. This paper aims to estimate the daily ground-level NO2 concentrations over China based on machine learning models and the TROPOMI NO2 data with high spatial resolution. In this study, four tree-based algorithm machine learning models, decision trees (DT), gradient boost decision tree (GBDT), random forest (RF) and extra-trees (ET), were used to estimate ground-level NO2 concentrations. In addition to considering many influencing factors of the ground-level NO2 concentrations, we especially introduced simplified temporal and spatial information into the estimation models. The results show that the extra-trees with spatial and temporal information (ST-ET) model has great performance in estimating ground-level NO2 concentrations with a cross-validation R2 of 0.81 and RMSE of 3.45 μg/m3 in test datasets. The estimated results for 2019 based on the ST-ET model achieves a satisfactory accuracy with a cross-validation R2 of 0.86 compared with the other models. Through time-space analysis and comparison, it was found that the estimated high-resolution results were consistent with the ground observed NO2 concentrations. Using data from January 2020 to test the prediction power of the models, the results indicate that the ST-ET model has a good performance in predicting ground-level NO2 concentrations. Taking four ground-level NO2 concentrations hotspots as examples, the estimated ground-level NO2 concentrations and ground-based observation data during the coronavirus disease (COVID-19) pandemic were lower compared with the same period in 2019. The findings offer a solid solution for accurately and efficiently estimating ground-level NO2 concentrations by using satellite observations, and provide useful information for improving our understanding of the regional atmospheric environment.

The impact of public health emergencies on hotel demand - Estimation from a new foresight perspective on the COVID-19.

This paper proposes a new foresight approach to estimate the impact of public health emergencies on hotel demand. The forecasting-based influence evaluation consists of four modules: decomposing hotel demand before an emergency, matching each decomposed component to a forecasting model, combining the predictions as the expected demand after the emergency, and estimating the impact by comparing actual demand against that predicted. The method is applied to analyze the impact of COVID-19 on Macao's hotel industry. The empirical results show that: 1) the new approach accurately estimates COVID-19's impact on hotel demand; 2) the seasonal and industry development components contribute significantly to the estimate of expected demand; 3) COVID-19's impact is heterogeneous across hotel services.

Ligand/Receptor-mediated Cell Communication Alteration Causes the Lung Inflammation Microenvironment and the Redistribution of Inter-tissue Immune Cells in COVID-19 Patients (preprint)

How SARS-CoV-2 causes disturbances of the lung microenvironment and systemic immune response remains a mystery. Here, we first analyze detailedly paired single-cell transcriptome data of the lungs, blood and bone marrow of two patients who died of COVID-19. Second, our results demonstrate that SARS-CoV-2 infection significantly increases the cellular communication frequency between AT1/AT2 cells and highly inflammatory myeloid cells, and induces the pulmonary inflammation microenvironment, and drives the disorder of fibroblasts, club and ciliated cells, thereby causing the increase of pulmonary fibrosis and mucus accumulation. Third, our works reveal that the increase of the lung T cell infiltration is mainly recruited by myeloid cells through certain ligands/receptors (ANXA1/FPR1, C5AR1/RPS19 and CCL5/CCR1), rather than AT1/AT2. Fourth, we find that some ligands and receptors such as ANXA1/FPR1, CD74/COPA, CXCLs/CXCRs, ALOX5/ALOX5AP, CCL5/CCR1, are significantly activated and shared among patients’ lungs, blood and bone marrow, implying that dysregulated ligands and receptors may cause the migration, redistribution and the inflammatory storm of immune cells in different tissues. Overall, our study reveals a latent mechanism by which the disorders of ligands and receptors caused by SARS-CoV-2 infection drive cell communication alteration, the pulmonary inflammatory microenvironment and systemic immune responses across tissues in COVID-19 patients.

Dual spatially resolved transcriptomics for SARS-CoV-2 host-pathogen colocalization studies in humans (preprint)

To advance our understanding of cellular host-pathogen interactions, technologies that facilitate the co-capture of both host and pathogen spatial transcriptome information are needed. Here, we present an approach to simultaneously capture host and pathogen spatial gene expression information from the same formalin-fixed paraffin embedded (FFPE) tissue section using the spatial transcriptomics technology. We applied the method to COVID-19 patient lung samples and enabled the dual detection of human and SARS-CoV-2 transcriptomes at 55 m resolution. We validated our spatial detection of SARS-CoV-2 and identified an average specificity of 94.92% in comparison to RNAScope and 82.20% in comparison to in situ sequencing (ISS). COVID-19 tissues showed an upregulation of host immune response, such as increased expression of inflammatory cytokines, lymphocyte and fibroblast markers. Our colocalization analysis revealed that SARS-CoV-2 + spots presented shifts in host RNA metabolism, autophagy, NF{kappa}B, and interferon response pathways. Future applications of our approach will enable new insights into host response to pathogen infection through the simultaneous, unbiased detection of two transcriptomes.

Ligand/Receptor-mediated Cell Communication Alteration Causes the Lung Inflammation Microenvironment and the Redistribution of Inter-tissue Immune Cells in COVID-19 Patients (preprint)

How SARS-CoV-2 causes disturbances of the lung microenvironment and systemic immune response remains a mystery. Here, we first analyze detailedly paired single-cell transcriptome data of the lungs, blood and bone marrow of two patients who died of COVID-19. Second, our results demonstrate that SARS-CoV-2 infection significantly increases the cellular communication frequency between AT1/AT2 cells and highly inflammatory myeloid cells, and induces the pulmonary inflammation microenvironment, and drives the disorder of fibroblasts, club and ciliated cells, thereby causing the increase of pulmonary fibrosis and mucus accumulation. Third, our works reveal that the increase of the lung T cell infiltration is mainly recruited by myeloid cells through certain ligands/receptors (ANXA1/FPR1, C5AR1/RPS19 and CCL5/CCR1), rather than AT1/AT2. Fourth, we find that some ligands and receptors such as ANXA1/FPR1, CD74/COPA, CXCLs/CXCRs, ALOX5/ALOX5AP, CCL5/CCR1, are significantly activated and shared among patients’ lungs, blood and bone marrow, implying that dysregulated ligands and receptors may cause the migration, redistribution and the inflammatory storm of immune cells in different tissues. Overall, our study reveals a latent mechanism by which the disorders of ligands and receptors caused by SARS-CoV-2 infection drive cell communication alteration, the pulmonary inflammatory microenvironment and systemic immune responses across tissues in COVID-19 patients.

Soundscape Preference of Urban Residents in China in the Post-pandemic Era

This research aims to explore the reality of the soundscape preferences of Chinese urban residents in general public landscape in the post-pandemic era, and then to propose design recommendations to meet the practical needs of people’s preferences for landscape—especially soundscapes—in the post-pandemic era. In this study, we utilized the subjective evaluation method to conduct an online questionnaire in 29 Chinese provinces which experienced severe pandemic caseloads and collected 860 valid responses. This study revealed people’s preference for landscape and soundscape in the post-pandemic era. We further studied the correlation between landscape preference and soundscape preference, analyzed the influence of living conditions on soundscape preference, founded the effects of personal characteristics and living conditions on soundscape preference, and explored the strongest influence factors on soundscape preference through the establishment of automatic linear model. The results revealed a positive correlation between life happiness and soundscape preference, whereas wearing masks significantly reduced soundscape perception ratings and people who have been vaccinated are more tolerant of various noises. Moreover, based on these analysis results, the design recommendations on landscape (overall landscape, plant, and tour space), soundscape construction of caring for vulnerable groups (teenagers and children, elderly people, and disabled and unhealthy) has been discussed.

An analysis of COVID-19 impacts on S&P 500 and FinTech index

COVID-19 developed into an extremely serious pandemic by the middle of 2020. It has caused enormous negative impacts such as a crush to the global market. In this study, we tested the correlation between COVID-19 and stock market in a more intuitive way with the COVID-19 transmission rate and recovery rate. They were generated by using Unscented Kalman Filter method incorporated with SEIR model. Since the Unscented Kalman Filter method analyzes data at daily intervals, we can study the trend of COVID-19 development and the fund index rate change in detail.

Clinical Efficacy of Convalescent Plasma Therapy on Treating COVID-19 Patients: Evidence from Matched Study and a Meta-Analysis (preprint)

Background: COVID-19 has infected tens of millions of people worldwide since its pandemic. CPT is one of the promising treatment methods and is favored by more and more researchers. However, the clinical efficacy and safety of CPT in COVID-19 remains unclear.Methods: We performed a matched control study by PSM analysis (including 163 cases with CPT and 163 controls with the standard treatment) and meta-analysis (including 498 cases and 557 controls) estimate the clinical efficacy and security of CPT and COVID-19, which will help inform clinical management of COVID-19 infection.Results: We found that days of hospital stay in case with CPT groups were significantly higher than matched control group (P< 0.0001). A significant reduction in mortality (OR= 0.496, 95%CI= 0.342-0.719, P< 0.0001) was found in the CPT group compared with the standard treatment group, and a true positive result was also found in sequential analysis. In terms of adverse events, sequential analysis found a false positive, although meta-analysis found a significant increase in the incidence of adverse events in patients treated with CPT compared to the control group. No differences between the two groups in terms of length of stay, improvement of clinical symptoms, and discharge were found.Conclusions: This study is the first to systematically review and meta-analysis the efficacy and safety of CPT in patients with COVID-19 in the largest sample size. Our results showed that CPT could significantly reduce the mortality rate of COVID-19 patients, and there was no significant increase in the incidence of adverse events. These data provide evidence favoring the efficacy and safety of CPT as a therapeutic agent in COVID-19 patients and provide comprehensive reference for COVID-19 treatment.Funding Statement: This work was supported by Scientific Research Project of Jiangsu Commission of Health (H2019065), Key Foundation of Wuhan Huoshenshan Hospital (2020[18]), Key Research & Development Program of Jiangsu Province (BE2018713), Medical Innovation Project of Logistics Service (18JS005).Declaration of Interests: The authors declare no conflicts of interest with this work.Ethics Approval Statement: The authors were approved by the ethics committee of Huoshenshan hospital, and were conducted in accordance with the tenets of the Declaration of Helsinki and its amendments. All participants provided written informed consent for the collection of samples and their subsequent analysis.

Aspergillus But Not Candida: Fatal Fungus Infection in COVID-19 Critical Patients (preprint)

Objectives: Earlier researches suggested patients should be routinely screened for bacteria and fungi infection after COVID-19 being confirmed. Here, we enrolled 236 patients with COVID-19 to analyze the clinical characteristics, fungal strains, mortality, and laboratory data of different groups.Design: Single center retrospective studyPatients: A total of 236 COVID-19 patients from Huoshenshan Hospital were included in this study, consisting of 14(6%) died cases, 222(94%) discharged cases.Results: The result revealed that 5 mortality in positive group were all related to aspergillus infection while candida infection rarely caused death. Aspergillus was most common in non-survivors while candida was most common in survivors. In terms of interleukin-6 (IL6), viral loads, nucleic acid clearance time, etc, fungal serologically positive group had a higher level than negative group.Conclusions: Non-survivors of Covid-19 with fungal infection were almost associated with aspergillus infection. Aspergillus infection, instead of candida infection might be fatal for critical ill patients with COVID-19. There is great significance to carry out routine screening for fungal infection especially for critical patients to enable early treatment to be implemented.Funding Statement: This study was financially supported by grants Key Foundation of Wuhan Huoshenshan Hospital (2020[18]), Key Research& Development Program of Jiangsu Province (BE2018713), Medical Innovation Project of Logistics Service (18JS005).Declaration of Interests: The authors declare no competing interests.Ethics Approval Statement: This study was approved by the Medical Ethical Committee of Wuhan Huoshenshan Hospital (No. HSSLL011). Written informed consent was obtained from each patient.

Diabetes Patients of COVID-19 Infection Treated with Convalescent Plasma Transfusion in Wuhan, China (preprint)

Background: The coronavirus disease 2019 (COVID-19) caused by SARS-CoV-2 had spread all over the world, causing public health emergency. Although the diagnosis for COVID-19 such as nucleic acid test and antibody detection have been well defined, there is still a big gap of knowledge regarding for COVID-19 patients receiving convalescent plasma transfusion (CPT) therapy, especially patients with comorbidity of diabetes. Method: In this study, out of 3059 COVID-19 patients admitted in Wuhan Huoshenshan Hospital of China, we described the characteristics of 39 diabetes patients receiving the transfusion of ABO-compatible convalescent plasma, and compared the baseline information and clinical outcome with that of 328 diabetes patients receiving traditional treatment. Results: It was found that the intervention of CPT therapy was effective and beneficial for COVID-19 patients, including severe or critical patients with comorbidity of diabetes, without obvious adverse effects observing during the treatments. The CPT therapy significantly improved the clinical outcome of diabetes patients with COVID-19 infection, especially the duration based on six categories compared to the patients with traditional therapy. Conclusions: This study not only provided a better understanding of COVID-19 in diabetes people receiving CPT, but also highlighted the CPT therapy was helpful for COVID-19 patients with comorbidity of diabetes.

Establishment of a clinical nomogram model to predict the progression of COVID-19 to severe disease (preprint)

Background Coronavirus disease 2019 (COVID-19) is a worldwide public health pandemic with a high mortality rate, among severe cases. The disease is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus. It is important to ensure early detection of the virus to curb disease progression to severe COVID-19. This study aimed to establish a clinical-nomogram model to predict the progression to severe COVID-19 in a timely, efficient manner.Methods This retrospective study included 202 patients with COVID-19 who were admitted to the Fifth Affiliated Hospital of Sun Yat-sen University and Shiyan Taihe Hospital from January 17 to April 30, 2020. The patients were randomly assigned to the training dataset (n = 163, with 43 progressing to severe COVID-19) or the validation dataset (n = 39, with 10 progressing to severe COVID-19) at a ratio of 8:2. The optimal subset algorithm was applied to filter for the clinical factors most relevant to the disease progression. Based on these factors, the logistic regression model was fit to distinguish severe (including severe and critical cases) from non-severe (including mild and moderate cases) COVID-19. Sensitivity, specificity, and area under the curve (AUC) were calculated using the R software package to evaluate prediction performance. A clinical nomogram was established and performance assessed with the discrimination curve.Results Risk factors, including demographics data, symptoms, laboratory and image findings were recorded for the 202 patients. Eight of the 52 variables that were entered into the selection process were selected via the best subset algorithm to establish the predictive model; they included gender, age, BMI, CRP, D-dimer, TP, ALB, and involved-lobe. Sensitivity, specificity and AUC were 0.91, 0.84 and 0.86 for the training dataset, and 0.87, 0.66, and 0.80 for the validation dataset.Conclusions We established an efficient and reliable clinical nomogram model which showed that gender, age, and initial indexes including BMI, CRP, D-dimer, involved-lobe, TP, and ALB could predict the risk of progression to severe COVID-19.

Impact of Coronavirus Disease 2019 on Pulmonary Function in Early Convalescence Phase (preprint)

Objective: This study investigated the influence of Coronavirus Disease 2019 (COVID-19) on lung function in early convalescence phase. Methods: A prospective retrospective study of COVID-19 patients at the Fifth Affiliated Hospital of Sun Yat-sen University were conducted, with serial assessments including lung volumes (TLC), spirometry (FVC, FEV1), lung diffusing capacity for carbon monoxide (DLCO),respiratory muscle strength, 6-minute walking distance (6MWD) and high resolution CT being collected at 30 days after discharged. Results: 57 patients completed the serial assessments. There were 40 non-severe cases and 17 severe cases. Thirty-one patients (54.3%) had abnormal CT findings. Abnormalities were detected in the pulmonary function tests in 43 (75.4%) of the patients. Six (10.5%), 5(8.7%), 25(43.8%) 7(12.3%), and 30 (52.6%) patients had FVC, FEV1, FEV1/FVC ratio, TLC, and DLCO values less than 80% of predicted values, respectively. 28 (49.1%) and 13 (22.8%) patients had PImax and PEmax values less than 80% of the corresponding predicted values. Compared with non-severe cases, severe patients showed higher incidence of DLCO impairment (75.6%vs42.5%, p=0.019), higher lung total severity score（TSS）and R20, and significantly lower percentage of predicted TLC and 6MWD. No significant correlation between TSS and pulmonary function parameters was found during follow-up visit. Conclusion: Impaired diffusing-capacityDeclining DLCO, lower respiratory muscle strength, and lung imaging abnormalities were detected in more than half of the COVID-19 patients in early convalescence phase. Compared with non-severe cases, severe patients had a higher incidence of DLCO impairment and encountered more TLC decrease and 6MWD decline.

Clinical characteristics and health care cost among patients successfully treated for COVID-19 in Henan, China: A Descriptive Study (preprint)

Objective: To clarify the clinical and medical expense characteristics of COVID-19. Methods: In this retrospective, single-center study, 55 cured cases with confirmed COVID-19 were analyzed for demographic, epidemiological, clinical, and radiological features and medical expense data.Results: The average age of the 54 successfully treated patients with COVID-19 was 53.2 years old (SD 19.0), including 27 men and 27 women. Off this, 31 (57.4%) patients had chronic diseases. Patients commonly had clinical manifestations of fever (45 [83.3%] patients), cough (29[54.7%] patients), expectoration (28 [51.9%] patients), fatigue (24[44.4%] patients) and diarrhea (8[14.8%] patients) on admission. There was a 10-day interval from the onset of signs and symptoms to hospital admission. About 80% of them got recovery after a two-week treatment. The mean interval from the onset of signs and symptoms to hospital discharge was 20.5 (IQR 16-29) days. The median total medical expense of the treated patient, in general, was 2579.6 (IQR 1366.1-4837.6) U.S. dollars. Still, the median medical expense was 8904.1 (IQR 6660.1- 27143.8) U.S. dollars in patients with more than five comorbid illnesses during the treatment.Conclusion: There is a 3-week interval from the onset of signs and symptoms to cure, and most hospitalized patients get recovery within two weeks. The total medical expense of cases with more than five comorbid conditions during the treatment is higher. Quite a few COVID-19 cases with other serious diseases are likely to account for most of the total medical expenses. 

The Dynamic Changes of Antibodies against SARS-CoV-2 during the Infection and Recovery of COVID-19 (preprint)

Deciphering the dynamic changes of antibodies against SARS-CoV-2 is essential for understanding the immune response in COVID-19 patients. By comprehensively analyzing the laboratory findings of 1,850 patients, we describe the dynamic changes of the total antibody, spike protein (S)-, receptor-binding domain (RBD)-, and nucleoprotein (N)- specific IgM and IgG levels during SARS-CoV-2 infection and recovery. Our results indicate that the S-, RBD-, and N- specific IgG generation of severe/critical COVID-19 patients is one week later than mild/moderate cases, while the levels of these antibodies are 1.5-fold higher in severe/critical patients during hospitalization (P<0.01). The decrease of these IgG levels indicates the poor outcome of severe/critical patients. The RBD- and S-specific IgG levels are 2-fold higher in virus-free patients (P<0.05). Notably, we found that the patients who got re-infected had a low level of protective antibody on discharge. Therefore, our evidence proves that the dynamic changes of antibodies could provide an important reference for diagnosis, monitoring, and treatment, and shed new light on the precise management of COVID-19.

Key to successful treatment of COVID-19: accurate identification of severe risks and early intervention of disease progression (preprint)

Abstract Background COVID-19 is a new and highly contagious respiratory disease that has caused global spread, high case fatality rate in severe patients, and a huge medical burden due to invasive mechanical ventilation. The current diagnosis and treatment guidelines are still need to be improved, and more excellent clinical experience is needed to provide reference. Methods We analyzed and summarized clinical data of 97 confirmed COVID-19 adult patients (including 26 severe cases) admitted to the Fifth Affiliated Hospital of Sun Yat-sen University from January 17, 2020 to March 10, 2020,included laboratory examination results, imaging findings, treatment effect, prognosis , etc, in order to put forward prediction index of severe COVID-19 patients, principles of early intervention and methylprednisolone usages in COVID-19 patients. Results 1.Hypoxemia, hyperlactic acid, hypoproteinemia, and hypokalemia were prevalent in COVID-19 patients.The significant low lymphocyte count, hypoproteinemia, hypokalemia, the persistent or worsen high CRP, high D-dimer, and high BNP, and the occurrence of hemoptysis and novel coronavirus (SARS-CoV-2) viremia were important indicators for early diagnosis and prediction of severe disease progression. 2.Characteristic images of lung CT had a clear change in COVID - 19, Ground-glass opacity (GGO) and high-density linear combinations may indicate different pathological changes. Rapid lobular progression of GGO suggests the possibility of severe disease. 3.Basic principles of early intervention treatment of COVID-19: on the premise of no effective antiviral drugs, treatment is based on supportive and symptomatic therapy (albumin supplementation, supplement of potassium, supplement blood plasma, etc.) in order to maintain the stability of the intracellular environment and adequately reactivate body immunity to clean up SARS-CoV-2 . 4. According to severity, oxygenation index, body weight, age, underlying diseases, appropriate amount methylprednisolone application on severe/critical COVID-19 patients on demand, improved blood oxygen and reduced the utilization rate of invasive mechanical ventilation, case fatality rate and medical burden significantly. The most common indications for invasive mechanical ventilation should be strictly control in critical COVID-19 patients. Conclusions: 1.Accurate and timely identification of clinical features in severe risks, and early and appropriate intervention can block disease progression. 2.Appropriate dose of methylprednisolone can effectively avoid invasive mechanical ventilation and reduce case fatality rate in critical COVID-19 patients.

Establishment of a clinical nomogram model to predict the progress to severe COVID-19 (preprint)

Background: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is the leading cause of a public health emergency in the world, accompanying with high mortality in severe corona virus disease 2019(COVID-19 ), thereby early detection and stopping the progress to severe COVID-19 is important. Our aim is to establish a clinical nomogram model to calculate and predict the progress to severe COVID-19 timely and efficiently.Methods: In this study, 65 patients with COVID-19 had been included retrospectively in the Fifth Affiliated Hospital of Sun Yat-sen University from January 17, to February 11, 2020. Patients were randomly assigned to train dataset (n=51 with 15 progressing to severe COVID-19) and test dataset (n=14 with 4 progressing to severe COVID-19). Lasso algorithm was applied to filter the most classification relevant clinical factors. Based on selected factors, logistic regression model was fit to predict the severe from mild/common. Meanwhile in nomogram sensitivity, specificity, AUC (Area under Curve), and calibration curve were depicted and calculated by R language, to evaluate the prediction performance to severe COVID-19.Results:High ratio of sever COVID-19 patients (26.5%) had been found in our retrospective study, and 84% of these cases progress to severe or critical after 5 days from their first clinical examination. In these 65 patients with COVID-19, 77 clinical characteristics in first examination were collected and analyzed, and 37 ones had been found different between non-severe and severe COVID-19. But when all these factors were analyzed in establishment of prediction model, six factors are crucial for predicting progress of severe COVID-19 via Lasso algorithm. Based on these six factors, including increased fibrinogen, hyponatremia, decreased PaO2,multiple lung lobes involved, down-regulated CD3（+）T-lymphocyte and fever, a logistic regression model was fit to discriminate severe and common COVID-19 patients. The sensitivity, specificity and AUC were 0.93, 0.86, 0.96 in the train dataset and 0.9, 1.0, 1.0 in test dataset respectively. Nomogram-predicted probability was more consistent with actual probability by R language.Conclusions:In summary, an efficient and reliable clinical nomogram model had been established, which indicate increased fibrinogen, hyponatremia, decreased PaO2, multiple lung lobes involved, down-regulated CD3（+）T-lymphocyte and fever at the first clinical examination, could predict progress of patients to severe COVID-19.