Fig. 2: Establishment and evaluation of ITG proteomic signature. | npj Aging

Fig. 2: Establishment and evaluation of ITG proteomic signature.

From: Unbiased proteomics and multivariable regularized regression techniques identify SMOC1, NOG, APCS, and NTN1 in an Alzheimer’s disease brain proteomic signature

Fig. 2

a Output of eNetXplorer model selection analyses, performed in ROS ITG. The x-axis indicates the model alpha value, a mixing parameter that scales continuously from the ridge (alpha = 0) to lasso (alpha = 1) regularized regression models. The red line indicates the out-of-bag prediction performance using accuracy as quality function (QF), which captures the proportion of class (AD vs. CN) predictions that match the true class of each participant. The blue line indicates the estimated model significance for each alpha value. b Contingency matrix comparing the out-of-bag predicted class and true class for the selected model (alpha = 1). c Receiver operating characteristic (ROC) curve for leave-one-out cross-validation (LOOCV) of the four protein ITG signature in the ROS cohort, derived at the alpha = 1 level. Gray shading indicates the 95 percent confidence interval. d ROC curve for LOOCV of ITG samples from an independent cohort of BLSA participants. The area under the curve (AUC) is demonstrated with 95 percent confidence intervals. Gray shading indicates the 95 percent confidence interval.

Back to article page