Fig. 5

The logistic regression model accurately predicts EC and benign conditions using protein biomarkers in CVL samples. The least absolute shrinkage and selection operator (LASSO) was performed to select features to build the logistic regression model (A). Twelve proteins with 100% frequency of LASSO selection were used to build the model. The performance of the model was evaluated using the Monte Carlo cross-validation. A multivariate ROC analysis, showing true and false positive rates, indicates excellent prediction of EC when compared to benign conditions (AUC 0.91) (B). A scatterplot depicts the predicted class probabilities of all samples using the classifier at a threshold of 0.5 (C). The confusion matrix illustrates the proportion of times each sample receives the correct classification (D). The logistic regression model correctly classified 151 out of 174 tested samples (86.8%)