Machine Learning Analysis of Image Data Based on Detailed MR Image Reports for Nasopharyngeal Carcinoma Prognosis

Table 3

The mean AUC of image data-based and TNM system-based ML model.

	OS	DMFS	LRFS
Image data (AUC)	0.796 (0.044)	0.752 (0.042)	0.721 (0.052)
TNM system (AUC)	0.712 (0.064)	0.693 (0.050)	0.617 (0.073)
value (AUC)	0.008	0.053	0.025
Image data (test error)	0.208 (0.037)	0.271 (0.052)	0.287 (0.050)
TNM system (test error)	0.326 (0.052)	0.346 (0.031)	0.413 (0.047)
value (test error)	0.006	0.011	0.006
Image data (specificity)	0.721 (0.061)	0.576 (0.114)	0.540 (0.153)
TNM system (specificity)	0.405 (0.132)	0.383 (0.060)	0.174 (0.010)
value (specificity)	0.006	0.011	0.006

The average performance of the two models is reported with standard deviation in the parenthesis.