Machine learning-driven blood transcriptome-based discovery of SARS-CoV-2 specific severity biomarkers

Machine learning versus multivariate logistic regression for predicting severe COVID-19 in hospitalized children with Omicron variant infection

With the emergence of the Omicron variant, the number of pediatric Coronavirus Disease 2019 (COVID-19) cases requiring hospitalization and developing severe or critical illness has significantly increased. Machine learning and multivariate logistic regression analysis were used to predict risk factors and develop prognostic models for severe COVID-19 in hospitalized children with the Omicron variant in this study. Of the 544 hospitalized children including 243 and 301 in the mild and severe groups, respectively. Fever (92.3%) was the most common symptom, followed by cough (79.4%), convulsions (36.8%), and vomiting (23.2%). The multivariate logistic regression analysis showed that age (1-3 years old, odds ratio (OR): 3.193, 95% confidence interval (CI): 1.778-5.733], comorbidity (OR: 1.993, 95% CI:1.154-3.443), cough (OR: 0.409, 95% CI:0.236-0.709), and baseline neutrophil-to-lymphocyte ratio (OR: 1.108, 95% CI: 1.023-1.200), lactate dehydrogenase (OR: 1.993, 95% CI: 1.154-3.443), blood urea nitrogen (OR: 1.002, 95% CI: 1.000-1.003) and total bilirubin (OR: 1.178, 95% CI: 1.005-3.381) were independent risk factors for severe COVID-19. The area under the curve (AUC) of the prediction models constructed by multivariate logistic regression analysis and machine learning (RandomForest + TomekLinks) were 0.7770 and 0.8590, respectively. The top 10 most important variables of random forest variables were selected to build a prediction model, with an AUC of 0.8210. Compared with multivariate logistic regression, machine learning models could more accurately predict severe COVID-19 in children with Omicron variant infection.

Biomarker integration for improved biodosimetry of mixed neutron + photon exposures

There is a persistent risk of a large-scale malicious or accidental exposure to ionizing radiation that may affect a large number of people. Exposure will consist of both a photon and neutron component, which will vary in magnitude between individuals and is likely to have profound impacts on radiation-induced diseases. To mitigate these potential disasters, there exists a need for novel biodosimetry approaches that can estimate the radiation dose absorbed by each person based on biofluid samples, and predict delayed effects. Integration of several radiation-responsive biomarker types (transcripts, metabolites, blood cell counts) by machine learning (ML) can improve biodosimetry. Here we integrated data from mice exposed to various neutron + photon mixtures, total 3 Gy dose, using multiple ML algorithms to select the strongest biomarker combinations and reconstruct radiation exposure magnitude and composition. We obtained promising results, such as receiver operating characteristic curve area of 0.904 (95% CI: 0.821, 0.969) for classifying samples exposed to ≥ 10% neutrons vs. < 10% neutrons, and R2 of 0.964 for reconstructing photon-equivalent dose (weighted by neutron relative biological effectiveness) for neutron + photon mixtures. These findings demonstrate the potential of combining various -omic biomarkers for novel biodosimetry.

Investigation of shared genes and regulatory mechanisms associated with coronavirus disease 2019 and ischemic stroke

Objective

Clinical associations between coronavirus disease (COVID-19) and ischemic stroke (IS) have been reported. This study aimed to investigate the shared genes between COVID-19 and IS and explore their regulatory mechanisms.

Methods

Published datasets for COVID-19 and IS were downloaded. Common differentially expressed genes (DEGs) in the two diseases were identified, followed by protein–protein interaction (PPI) network analysis. Moreover, overlapping module genes associated with the two diseases were investigated using weighted correlation network analysis (WGCNA). Through intersection analysis of PPI cluster genes and overlapping module genes, hub-shared genes associated with the two diseases were obtained, followed by functional enrichment analysis and external dataset validation. Moreover, the upstream miRNAs and transcription factors (TFs) of the hub-shared genes were predicted.

Results

A total of 91 common DEGs were identified from the clusters of the PPI network, and 129 overlapping module genes were screened using WGCNA. Based on further intersection analysis, four hub-shared genes in IS and COVID-19 were identified, including PDE5A, ITGB3, CEACAM8, and BPI. These hub-shared genes were remarkably enriched in pathways such as ECM-receptor interaction and focal adhesion pathways. Moreover, ITGB3, PDE5A, and CEACAM8 were targeted by 53, 32, and 3 miRNAs, respectively, and these miRNAs were also enriched in the aforementioned pathways. Furthermore, TFs, such as lactoferrin, demonstrated a stronger predicted correlation with the hub-shared genes.

Conclusion

The four identified hub-shared genes may participate in crucial mechanisms underlying both COVID-19 and IS and may exhibit the potential to be biomarkers or therapeutic targets for the two diseases.