Enhancing prediction of supraspinatus/infraspinatus tendon complex injuries through integration of deep visual features and clinical information: a multicenter two-round assessment study

Insights into Imaging

Table 3 Diagnostic performance of the deep learning algorithm

Models	Sensitivity	Specificity	Accuracy	AUC	95% CI
CML models
Lasso	0.893	0.667	0.763	0.832	0.773–0.890
SVM	0.933	0.627	0.757	0.866	0.815–0.917
Decision tree	0.759	0.753	0.756	0.826	0.795–0.856
Random forest	0.783	0.863	0.829	0.897	0.8750–0.919
DL models
Resnet-101	0.793	0.596	0.678	0.753	0.717–0.788
VGG-19	0.827	0.637	0.718	0.788	0.721–0.854
Inception-V3	0.793	0.596	0.678	0.753	0.717–0.788
Ensemble DL model	0.8	0.653	0.712	0.797	0.734–0.861
Ensemble CML-DL model	0.88	0.812	0.836	0.902	0.858–0.947