Skip to main content
Fig. 1 | Journal for ImmunoTherapy of Cancer

Fig. 1

From: Development of a prognostic composite cytokine signature based on the correlation with nivolumab clearance: translational PK/PD analysis in patients with renal cell carcinoma

Fig. 1

a Schematic overview of the machine-learning approach used to identify and then validate the composite prognostic biomarkers. b AUC-ROC analysis to show the performance of the machine-learning model (AUC = 0.7). c 2 × 2 analysis for actual clearance vs predicted clearance to show the accuracy of the model performance. d Selected cytokine features from the machine-learning model based on measured importance. Eight top-ranking cytokines were selected to form a composite signature: C-reactive protein (CRP), ferritin (FRTN), tissue inhibitor of metalloproteinase 1 (TIMP-1), brain-derived neurotrophic factor (BDNF), alpha 2-macroglobulin (A2Macro), stem cell factor (SCF), vascular endothelial growth factor-3 (VEGF-3), and intercellular adhesion molecule 1 (ICAM-1). AUC-ROC area under the receiver operating characteristic curve, CL clearance, F1 harmonic mean of precision and recall, NIVO nivolumab

Back to article page