Figure 7.

Verification and comparison of the decision tree model in predicting low and high grade renal cell carcinoma in the training and testing set. In the training set, the true negative rate (specificity) was 77.8% (63/81) for predicting grade I-II clear cell renal cell carcinoma (ccRCC), and the true positive rate (sensitivity) was 55.8% (29/52). In the testing set, the true negative rate was 82.9% (34/41) for predicting grade I-II ccRCC and 62.5% (10/16) for predicting grade III-IV ccRCC.