Identifying diagnostic feature biomarkers using multiple machine learning algorithms. (A) The scale-free network analysis for various soft-thresholding power (β). (B) Heatmap shows the correlation between the module eigengenes and clinical characteristics of the disease. (C and D) The "LASSO" algorithm shows the optimal coefficient and minimal lambda of DE-MRGs. (E) The "SVM" algorithm shows the highest accuracy of DE-MRGs. (F) The "RF" algorithm shows diagnostic feature biomarkers based on DE-MRGs with the" MeanDecreaseGini > 1." DE-MRGs = differentially expressed metabolism-related genes, LASSO = least absolute shrinkage and selection operator, RF = random forest, SVM = support vector machine.

Contexts in source publication

Context 1

... machine learning algorithms to screen HP's diagnostic feature biomarkers. WGCNA was used for constructing a gene co-expression network. The soft-thresholding parameter was set as "power of B = 5 (scale-free R 2 > 0.9) to establish the scale-free network." The module eigengenes' correlation coefficient and disease characteristics were calculated (Fig. 4A). The module trait results showed a correlation between the module eigengenes and disease characteristics. A positive correlation was observed between the module turquoise and HP, which was selected for further analysis (Fig. 4B). The genes of the module turquoise were shown in Table S2, Supplemental Digital Content, ...

Context 2

... the scale-free network." The module eigengenes' correlation coefficient and disease characteristics were calculated (Fig. 4A). The module trait results showed a correlation between the module eigengenes and disease characteristics. A positive correlation was observed between the module turquoise and HP, which was selected for further analysis (Fig. 4B). The genes of the module turquoise were shown in Table S2, Supplemental Digital Content, http://links.lww.com/MD/J737. Furthermore, we identified 25 DE-MRGs as diagnostic feature biomarkers for HP using the "LASSO" algorithm ( Fig. 4C and D). We identified 10 diagnostic feature biomarkers from DE-MRGs using the "SVM" algorithm (Fig. ...

Context 3

... positive correlation was observed between the module turquoise and HP, which was selected for further analysis (Fig. 4B). The genes of the module turquoise were shown in Table S2, Supplemental Digital Content, http://links.lww.com/MD/J737. Furthermore, we identified 25 DE-MRGs as diagnostic feature biomarkers for HP using the "LASSO" algorithm ( Fig. 4C and D). We identified 10 diagnostic feature biomarkers from DE-MRGs using the "SVM" algorithm (Fig. 4E). The "RF" algorithm was also used to identify 30 diagnostic feature biomarkers from DE-MRGs (Fig. ...

Context 4

... analysis (Fig. 4B). The genes of the module turquoise were shown in Table S2, Supplemental Digital Content, http://links.lww.com/MD/J737. Furthermore, we identified 25 DE-MRGs as diagnostic feature biomarkers for HP using the "LASSO" algorithm ( Fig. 4C and D). We identified 10 diagnostic feature biomarkers from DE-MRGs using the "SVM" algorithm (Fig. 4E). The "RF" algorithm was also used to identify 30 diagnostic feature biomarkers from DE-MRGs (Fig. ...

Context 5

... http://links.lww.com/MD/J737. Furthermore, we identified 25 DE-MRGs as diagnostic feature biomarkers for HP using the "LASSO" algorithm ( Fig. 4C and D). We identified 10 diagnostic feature biomarkers from DE-MRGs using the "SVM" algorithm (Fig. 4E). The "RF" algorithm was also used to identify 30 diagnostic feature biomarkers from DE-MRGs (Fig. ...