Fig. 2 | Scientific Reports

Fig. 2

From: Exploring exosome-related genes as candidate biomarkers in primary immune thrombocytopenia through transcriptomics and preliminary experimental validation

Fig. 2The alternative text for this image may have been generated using AI.Fig. 2The alternative text for this image may have been generated using AI.

(A) The Boruta algorithm was employed to perform feature selection on the 23 differentially expressed ERGs, ultimately identifying 5 feature genes (highlighted in green). (B–C) Screening of the 23 differentially expressed ERGs by LASSO regression selected 5 feature genes. (D)The intersection of two machine learning algorithms revealed 4 shared genes (Venn diagram). (E–G)The resulting nomogram was validated with calibration curves and ROC analysis, indicating its diagnostic efficacy. The discriminatory power of the model was assessed by the AUC, where numerical values closer to 1 signify superior model accuracy and reliability.

Back to article page