Receiver Operator Characteristic (ROC) Curves for coronary artery disease prediction.

<p>ROC curves plotted for the results of leave-one-out cross validation using logistic regression with only age (dashed), age and CAD (solid), and all variables (dotted) included in classification. Including CAD in classification yields significantly better results (p = 0.04) than only classifying on age. Including all variables yielded the highest area under the curve, but was not significantly better than classifying with only age and CAD (p = 0.28).