Development and External Validation of Machine Learning-Based Models for Predicting Lung Metastasis in Kidney Cancer: A Large Population-Based Study

Table 4

Predictive performance of algorithms in internal and external test cohorts.

Models	Internal test							External test
	AUC	Accuracy	Sensitivity	Specificity	Precision	Recall	F1 score	AUC	Accuracy	Sensitivity	Specificity	Precision	Recall	F1 score
LR	0.903	0.787	0.873	0.783	0.180	0.873	0.299	0.896	0.866	0.75	0.872	0.231	0.75	0.353
RF	0.791	0.885	0.661	0.897	0.260	0.661	0.373	0.703	0.876	0.5	0.895	0.197	0.5	0.283
SVM	0.61	0.938	0.326	0.972	0.389	0.326	0.355	0.674	0.919	0.167	0.957	0.167	0.167	0.167
XGB	0.913	0.812	0.873	0.809	0.200	0.873	0.325	0.904	0.872	0.75	0.878	0.240	0.75	0.364
DT	0.820	0.831	0.763	0.835	0.202	0.763	0.319	0.690	0.896	0.458	0.919	0.224	0.458	0.301
ANN	0.906	0.792	0.875	0.787	0.184	0.875	0.304	0.891	0.843	0.792	0.846	0.209	0.792	0.331

LR: logistic regression, RF: random forest, SVM: supporting vector machine, XGB: extreme gradient boosting, DT: decision tree, and ANN: artificial neural network.