Table 2.

Performances of the three deep learning models and three existing clinical scores for hemorrhagic transformation prediction in the test cohort

Models	AUROC	Sensitivity	Specificity
Clinical model	0.887 (0.864–0.910)	0.659	0.850
Imaging model	0.920 (0.897–0.945)	0.854	0.883
Ensemble model	0.937 (0.908–0.955)	0.878	0.883
MSS	0.707 (0.665–0.739)	0.659	0.733
GRASPS	0.791 (0.747–0.824)	0.585	0.917
SEDAN	0.824 (0.786–0.846)	0.854	0.633

Ensemble model is the integration of clinical and imaging models developed in this study. MSS, SEDAN, and GRASPS scores are three existing clinical scores for predicting hemorrhagic transformation of acute ischemic stroke

AUROC Area under the receiver operating characteristic curve

The bold font indicates that the AUC value of the Ensemble model is the highest