A prediction model for xerostomia in locoregionally advanced nasopharyngeal carcinoma patients receiving radical radiotherapy

Table 2 The predictive values of the models

	Sensitivity (95%CI)	Specificity (95%CI)	PPV (95%CI)	NPV (95%CI)	AUC (95%CI)	Accuracy (95%CI)
Training set
RF	1.000 (1.000–1.000)	0.968 (0.925–1.000)	0.990 (0.975–1.000)	1.000 (1.000–1.000)	0.999 (0.997–1.000)	0.992 (0.981–1.000)
XGB	0.974 (0.951–0.996)*	0.968 (0.925–1.000)	0.989 (0.975–1.000)	0.924 (0.860–0.988)*	0.995 (0.989–1.000)	0.973 (0.952–0.993)
DTC	0.943 (0.910–0.976)	0.984 (0.953–1.000)	0.995 (0.984–1.000)	0.849 (0.767–0.931)*	0.963 (0.941–0.986)*	0.953 (0.927–0.979)*
Testing set
RF	0.933 (0.880–0.985)	0.714 (0.521–0.908)	0.933 (0.880–0.985)	0.714 (0.521–0.908)	0.915 (0.860–0.970)	0.891 (0.833–0.949)
XGB	0.820 (0.740–0.900)	0.714 (0.521–0.908)	0.924 (0.866–0.982)	0.484 (0.308–0.660)	0.834 (0.753–0.916)	0.800 (0.725–0.875)
DTC	0.775 (0.689–0.862)^#	0.762 (0.580–0.944)	0.932 (0.875–0.990)^#	0.444 (0.282–0.607)^#	0.769 (0.666–0.872)^#	0.773 (0.694–0.851)^#

RF random forest, DTC decision tree classifier, XGB extreme-gradient boosting, PPV positive predictive value, NPV negative predictive value, AUC area under the curve
^*Compared with the training set in the RF model, the difference was statistically different
^#Compared with the testing set in the RF model, the difference was statistically different

ISSN: 1472-6831