Journal: Diagnostics

Article Title: Investigation of Long-Term CD4+ T Cell Receptor Repertoire Changes Following SARS-CoV-2 Infection in Patients with Different Severities of Disease

doi: 10.3390/diagnostics14202330

Figure Lengend Snippet: Exploratory analysis by PCA comparing COVID-19 patients (all severity groups; inclusive of all collected timepoints). ( a ) ANOVA test for each TCR vb family ( p < 0.05 is considered statistically significant [circled]). ( b ) PCA performed using the TCR vb families with p < 0.05. Clustering separation between the groups can be observed. ( c ) Key loadings (features) on PC1. ( d ) Key loadings on PC2. ( e ) Classification modelling using backpropagation artificial neural networks (ANNs); the model shows good separation for MA and IP critical, and lower performance is noted for IP severe followed by IP moderate (correct responses shaded in yellow).

Article Snippet: Multivariate analyses were performed by means of principal component analysis (PCA) and backpropagation artificial neural networks (ANNs) within the MATLAB environment using the PLS Toolbox version 7.9.3 (Eigenvector Research, Inc., Manson, WA, USA) and the Classification Toolbox by Ballabio and Consonni [ ].

Techniques:

Journal: Diagnostics

Article Title: Investigation of Long-Term CD4+ T Cell Receptor Repertoire Changes Following SARS-CoV-2 Infection in Patients with Different Severities of Disease

doi: 10.3390/diagnostics14202330

Figure Lengend Snippet: Exploratory analysis by PCA comparing COVID-19 IP severity groups (all timepoints). ( a ) ANOVA test for each TCR vb family ( p < 0.05 is considered statistically significant [circled]). ( b ) PCA performed using the TCR vb families with p < 0.05. Good clustering separation between the groups can be observed. ( c ) Key loadings (features) on PC1. ( d ) Classification modelling using backpropagation artificial neural networks (ANNs); good classification between IP severity groups can be obtained (correct responses shaded in yellow), thus, given the TCR VB family CD4+ percentage, the model can predict if the patient is moderate, severe, or critical. ( e ) ROC analysis curves; performance characteristics indicate outstanding discrimination for classification of moderate and critical disease IPs, and excellent discrimination for severe disease IPs, sensitivity (blue), specificity (red).

Article Snippet: Multivariate analyses were performed by means of principal component analysis (PCA) and backpropagation artificial neural networks (ANNs) within the MATLAB environment using the PLS Toolbox version 7.9.3 (Eigenvector Research, Inc., Manson, WA, USA) and the Classification Toolbox by Ballabio and Consonni [ ].

Techniques: