Figure 2
ROC-curve for predictions from XGBoost and the logistic regression model. The sensitivity and specificity of the one class SVM and clinical prediction rule are also plotted on the left curve. On the left the points corresponding to the 0.001 (‘t = 0.001’) and 0.002 (‘t = 0.002’) probability thresholds are plotted for the XGBoost and logistic regression model. On the right the points corresponding to the thresholds resulting in 138 positive predictions (‘t for 138 pos pred’, equaling the clinical rule positive predictions) are plotted for the XGBoost and logistic regression model.
