Development and validation of prediction models for papillary thyroid cancer structural recurrence using machine learning approaches

Table 2 Predictive performance of the models in the test dataset

	AUC (95% CI)	Sensitivity	Specificity	Accuracy	PPV	NPV	F1 score
LR	0.738 (0.636–0.820)	0.865	0.495	0.526	0.133	0.976	0.231
SVM	0.752 (0.666–0.841)	0.568	0.903	0.875	0.344	0.959	0.429
XGBoost	0.741 (0.609–0.840)	0.595	0.857	0.835	0.272	0.959	0.373
RF	0.766 (0.702–0.845)	0.676	0.784	0.775	0.219	0.964	0.331
NN	0.767 (0.675–0.843)	0.757	0.682	0.688	0.176	0.969	0.286
The ATA risk stratification	0.620 (0.534–0.670)	0.432	0.770	0.742	0.144	0.938	0.216

Abbreviations: AUC, area under the curve; PPV, positive predictive value; NPV, negative predictive value; LR, logistic regression; SVM, support vector machine; XGBoost, eXtreme gradient boosting; RF, random forest; NN, neural network; ATA, American Thyroid Association

ISSN: 1471-2407