Fig 1: Diagnostic performance evaluation of four candidate biomarkers (DUSP3, DSTN, PDIA5, CNST) through ROC curve analysis based on proteomics data. Diagnostic efficacy was assessed using ROC curves, and the AUC was calculated. A Individual biomarker ROC curves for distinguishing all dengue cases (DF + DFN) from HC. B Combined logistic regression model integrating all four biomarkers for DF + DFN vs HC. C Individual biomarker ROC curves for classic DF diagnosis after excluding DFN cases. D Multi-marker logistic regression model performance for DF vs HC. E Biomarker-specific ROC curves for DFN subgroup identification. F Integrated logistic regression model performance for DFN vs HC comparison. G Four biomarker-specific ROC curves for DFN vs. DF subgroup identification. H Integrated logistic regression model performance for DFN vs DF comparison.
Fig 2: Machine learning-based identification of key biomarkers for dengue diagnosis from 55 DEPs. Candidate biomarkers were screened using four machine learning algorithms (Random Forest, LASSO, Ridge, and Elastic Net). For validation, diluted plasma samples were incubated with specific antibodies and detected spectrophotometrically at 450 nm using ELISA to quantify protein concentrations. A Feature importance ranking obtained using Random Forest. B LASSO regression results showing the trajectory of coefficients, with the optimal lambda determined by minimizing the binomial deviance, thereby selecting the most significant proteins. C Ridge regression coefficients generated under L2 regularization. D Elastic Net regression coefficients, which integrate the properties of both L1 (LASSO) and L2 (Ridge) penalties for robust feature selection. E Venn diagram depicting the overlap of features selected by the four algorithms, converging on four key biomarkers (CNST, DSTN, DUSP3, and PDIA5) predictive of dengue. F Box plots illustrating the differential expression patterns of the four identified biomarkers (CNST, DSTN, DUSP3, and PDIA5) across HC, DF, and DFN groups. Statistical differences between any two groups were assessed using the independent samples t-test. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; and ∗∗∗P < 0.0001. HC: healthy controls. DF: dengue fever. DFN: dengue fever with neutropenia. DEPs: differentially expressed proteins. LASSO: least absolute shrinkage and selection operator.
Fig 3: The mean plasma levels of DSTN (A), PDIA5 (B), CNST (C) and DUSP3 (D) in patients from the HC, DF, and DFN groups as measured by ELISA. Bars represent mean ± SD. Statistical comparisons between groups were performed using one-way ANOVA followed by Tukey’s post-hoc test for multiple comparisons. ∗∗∗∗P < 0.0001.
Fig 4: Diagnostic performance evaluation of four candidate biomarkers (DUSP3, DSTN, PDIA5, CNST) through ROC curve analysis based on ELISA data. Diagnostic efficacy was assessed using ROC curves, and the AUC was calculated. A Individual biomarker ROC curves for distinguishing all dengue cases (DF + DFN) from HC. B Combined logistic regression model integrating all four biomarkers for DF + DFN vs HC. C Individual biomarker ROC curves for classic DF diagnosis after excluding DFN cases. D Multi-marker logistic regression model performance for DF vs HC. E Biomarker-specific ROC curves for DFN subgroup identification. F Integrated logistic regression model performance for DFN vs HC comparison. G Four biomarker-specific ROC curves for DFN vs. DF subgroup identification. F Integrated logistic regression model performance for DFN vs DF comparison.
Supplier Page from Fine Biotech Co., Ltd. for Human CNST (Consortin) ELISA Kit