Machine Learning-Based Scoring System for Early Prognosis Evaluation of Patients with Coronavirus Disease 2019

Background The global spread of coronavirus disease 2019 (COVID-19) continues to threaten human health security, exerting considerable pressure on healthcare systems worldwide. While prognostic models for COVID-19 hospitalized or intensive care patients are currently available, prognostic models developed for large cohorts of thousands of individuals are still lacking. Methods Between February 4 and April 16, 2020, we enrolled 3,974 patients admitted with COVID-19 disease in the Wuhan Huo-Shen-Shan Hospital and the Maternal and Child Hospital, Hubei Province, China. (1) Screening of key prognostic factors: A univariate Cox regression analysis was performed on 2,649 patients in the training set, and factors affecting prognosis were initially screened. Subsequently, a random survival forest model was established through machine analysis to further screen for factors that are important for prognosis. Finally, multivariate Cox regression analysis was used to determine the synergy among various factors related to prognosis. (2) Establishment of a scoring system: The nomogram algorithm established a COVID-19 patient death risk assessment scoring system for the nine selected key prognostic factors, calculated the C index, drew calibration curves and drew training set patient survival curves. (3) Verification of the scoring system: The scoring system assessed 1,325 patients in the test set, splitting them into high- and low-risk groups, calculated the C-index, and drew calibration and survival curves. Results The cross-sectional study found that age, clinical classification, sex, pulmonary insufficiency, hypoproteinemia, and four other factors (underlying diseases: blood diseases, malignant tumor;complications: digestive tract bleeding, heart dysfunction) have important significance for the prognosis of the enrolled patients with COVID-19. Herein, we report the discovery of the effects of hypoproteinemia and hematological diseases on the prognosis of COVID-19. Meanwhile, the scoring system established here can effectively evaluate objective scores for the early prognoses of patients with COVID-19 and can divide them into high- and low-risk groups (using a scoring threshold of 117.77, a score below which is considered low risk). The efficacy of the system was better than that of clinical classification using the current COVID-19 guidelines (C indexes, 0.95 vs. 0.89). Conclusions Age, clinical typing, sex, pulmonary insufficiency, hypoproteinemia, and four other factors were important for COVID-19 survival. Compared with general statistical methods, this method can quickly and accurately screen out the relevant factors affecting prognosis, provide an order of importance, and establish a scoring system based on the nomogram model, which is of great clinical significance. © Wolters Kluwer Health, Inc. All rights reserved.

[Expressions of MPV, P-LCR and NLR in patients with novel coronavirus disease 2019].

To provide new ideas for clinical diagnosis and treatment of coronavirus disease 2019 (COVID-19), this study explore the expression level and prognostic value of platelet parameters in mild, moderate and severe COVID-19. This is a retrospective analysis. From January to May 2020, a total of 69 patients who were diagnosed with COVID-19 in the Third Central Hospital and the Jinnan Hospital (both situated in Tianjin) were enrolled in the disease group. According to the severity, these patients were divided into mild group (15 cases), moderate group (46 cases), and severe group (8 cases). In the same period, 70 non-infected patients were enrolled in control group. The level of white blood cell count (WBC), absolute neutrophil count (NEU#), absolute lymphocyte count (LY#), neutrophil-lymphocyte ratio (NLR), red blood cell count (RBC), hemoglobin (Hb), platelet count (PLT), mean platelet volume (MPV), platelet distribution width (PDW), and platelet-large contrast ratio (P-LCR) before and after treatment were analyzed. Binary logistic regression analysis is used to establish a mathematical model of the relationship between these indexes and the outcome of severe COVID-19 patients. The receiver operating characteristic(ROC) curve is used to further explore the prognosis value of MPV, P-LCR, NLR separately and jointly in COVID-19 patients. Compare to the control group, WBC and NE# increase (Z=-5.63, P<0.01;Z=-9.19,P<0.01) and LY# decrease (Z=-9.34, P<0.01) in the severe group; NLR increase with the aggravation of the disease, there is significant difference between groups (Z=17.61, P<0.01); PLT, PDW, MPV and P-LCR decrease with the aggravation of the disease, there is significant difference between groups (Z=9.47, P<0.01; Z=11.41, P<0.01; Z =16.76, P<0.01; Z=13.97, P<0.01). Binary logistic regression analysis shows MPV, P-LCR and NLR have predictive value for severe COVID-19 patients. There is a negative correlation between MPV, P-LCR and severe COVID-19 patients (OR=1.004, P=0.034; OR=1.097, P=0.046). There is a positive correlation between NLR and severe COVID-19 patients (OR=1.052, P=0.016). MPV and P-LCR of patients with good prognosis after treatment were significantly higher than those before treatment (Z=-6.47, P<0.01; Z=-5.36, P<0.01). NLR was significantly lower than that before treatment (Z=-8.13, P<0.01). MPV and P-LCR in poor prognosis group were significantly lower than those before treatment (Z=-9.46, P<0.01; Z=-6.81, P<0.01). NLR was significantly higher than that before treatment (Z=-3.24, P<0.01). There were significant differences between good and poor prognosis groups before and after treatment in MPV, P-LCR and NLR (P<0.01). Combination of these three indexes, ROC shows the AUC is 0.931, the sensitivity is 91.5%, the specificity is 94.1%, the positive predictive value is 88.9%, and the negative predictive value is 87.4%, which is better than any of these indexes separately. Changes in these parameters are closely related to clinical stage of COVID-19 patients. MPV, P-LCR and NLR are of great value in the prediction and prognosis of severe COVID-19 patients.

Characteristic analysis and comparison of glycolipid metabolism in patients with coronavirus disease 2019 in common condition and severe cases

Objective•To analyze and compare the characteristics of glycolipid metabolism between common and severe patients with coronavirus disease 2019 (COVID-19). Methods•Thirty-six patients with COVID-19 were hospitalized in the general ward of Wuhan Leishenshan Hospital and fifty severe patients with COVID-19 in intensive care unit (ICU) from February to March, 2020. All the patients were divided into two groups: the common patient group and the severe patient group. Their electronic medical records were extracted and analyzed. The demographic data as well as clinical data, laboratory results, comorbidities and clinical outcomes in the two groups were collected and compared by independent sample t test, non-parametric test as well as χ2 test. From the metabolic point of view, the characteristics of glucose and lipid metabolism in COVID-19 common and severe patients and the possible related factors for patients staying in ICU were analyzed. Results•There was no significant difference between the two groups in terms of gender, number of patients with diabetes and coronary heart disease (CAD). The average age of severe patients was significantly older than that of the common patients (P<0.05). The proportion of the severe patients with hypertension (52.0%) was significantly higher than that of the common patients (22.2%) (P<0.05). The lymphocyte count of the severe patients was significantly lower than that of the common patients (P<0.05). There was no significant difference in glutamic-pyruvic transaminase (GPT), glutamic-oxaloacetic transaminase (GOT), serum creatinine (Scr) and blood uric acid (BUA) between the two groups. Blood serum albumin (ALB), adjusted calcium concentration (Cac), total cholesterol (TC), triacylglycerols (TAG), high density lipoprotein (HDL) and the low density lipoprotein (LDL) in the severe patients were significantly lower than those in the common patients (all P<0.05). Fasting blood glucose (FBG) in the severe patients was significantly higher than that in the common patients (P=0.001). Multivariate Logistic regression showed that the increase of FBG and the decrease of TC, HDL, LDL, ALB were related to COVID-19 patients staying in ICU. Conclusion•There are deteriorative disorders in terms of glucose and lipid metabolism among the severe patients with COVID-19. The FBG, TC, HDL, LDL and ALB may related to the admission of ICU.

[Exploration of omics mechanism and drug prediction of coronavirus-induced heart failure based on clinical bioinformatics].

Objective: Present study investigated the mechanism of heart failure associated with coronavirus infection and predicted potential effective therapeutic drugs against heart failure associated with coronavirus infection. Methods: Coronavirus and heart failure were searched in the Gene Expression Omnibus (GEO) and omics data were selected to meet experimental requirements. Differentially expressed genes were analyzed using the Limma package in R language to screen for differentially expressed genes. The two sets of differential genes were introduced into the R language cluster Profiler package for gene ontology (GO) and Kyoto gene and genome encyclopedia (KEGG) pathway enrichment analysis. Two sets of intersections were taken. A protein interaction network was constructed for all differentially expressed genes using STRING database and core genes were screened. Finally, the apparently accurate treatment prediction platform (EpiMed) independently developed by the team was used to predict the therapeutic drug. Results: The GSE59185 coronavirus data set was searched and screened in the GEO database, and divided into wt group, ΔE group, Δ3 group, Δ5 group according to different subtypes, and compared with control group. After the difference analysis, 191 up-regulated genes and 18 down-regulated genes were defined. The GEO126062 heart failure data set was retrieved and screened from the GEO database. A total of 495 differentially expressed genes were screened, of which 165 were up-regulated and 330 were down-regulated. Correlation analysis of differentially expressed genes between coronavirus and heart failure was performed. After cross processing, there were 20 GO entries, which were mainly enriched in virus response, virus defense response, type Ⅰ interferon response, γ interferon regulation, innate immune response regulation, negative regulation of virus life cycle, replication regulation of viral genome, etc. There were 5 KEGG pathways, mainly interacting with tumor necrosis factor (TNF) signaling pathway, interleukin (IL)-17 signaling pathway, cytokine and receptor interaction, Toll-like receptor signaling pathway, human giant cells viral infection related. All differentially expressed genes were introduced into the STRING online analysis website for protein interaction network analysis, and core genes such as signal transducer and activator of transcription 3, IL-10, IL17, TNF, interferon regulatory factor 9, 2'-5'-oligoadenylate synthetase 1, mitogen-activated protein kinase 3, radical s-adenosyl methionine domain containing 2, c-x-c motif chemokine ligand 10, caspase 3 and other genes were screened. The drugs predicted by EpiMed's apparent precision treatment prediction platform for disease-drug association analysis were mainly TNF-α inhibitors, resveratrol, ritonavir, paeony, retinoic acid, forsythia, and houttuynia cordata. Conclusions: The abnormal activation of multiple inflammatory pathways may be the cause of heart failure in patients after coronavirus infection. Resveratrol, ritonavir, retinoic acid, amaranth, forsythia, houttuynia may have therapeutic effects. Future basic and clinical research is warranted to validate present results and hypothesis.