Figure 4.
Analysis of the predictive model. (A) Box plot for the predictive score discriminating BPH from PCa. Data are shown as the sum of the variable levels multiplied by the regression coefficients as reported in panel C. (B) Contingency table corresponding to the cutoff of 21.34 for the predictive score. (C) Coefficients of the logistic regression for the selected variables. (D) ROC curve of the score obtained by combining age, miR-5100, and four metabolites after penalized logistic regression. BPH, benign prostatic hyperplasia; PCa, prostate cancer.
