Table 2.

Prehospital stroke prediction using machine learning.

Models	AUROC	Accuracy	Sensitivity	Specificity	F1-score
Training cohort
XGBoost	0.994	0.978	0.990	0.947	0.985
Random forest	0.979	0.943	0.956	0.910	0.960
SVM (Radial basis function)	0.968	0.928	0.950	0.873	0.950
SVM (Linear)	0.889	0.835	0.915	0.627	0.889
Logistic regression	0.882	0.843	0.847	0.835	0.886
Test cohort
XGBoost	0.980	0.952	0.986	0.864	0.967
Random forest	0.953	0.907	0.933	0.840	0.935
SVM (Radial Basis function)	0.935	0.900	0.933	0.815	0.931
SVM (Linear)	0.904	0.862	0.928	0.691	0.907
Logistic regression	0.886	0.828	0.828	0.827	0.874

AUROC area under the receiver operating characteristic curve, XGBoost eXtreme gradient boosting, SVM support vector machine.