Table 3.

Mean ROC-AUC value table of machine learning model versus prognostic outcomes (bold text indicates the highest value for each outcome or row; standard deviation values are in parentheses). As in Table 2, the p-value for each row’s best mean compares against the second-best performing model in the row

	Radiomics-based ANN	Lasso regression	Elastic regression	Logistic regression		Random forest	Support vector machine
Presence of IDRF	0.76 (0.021) (p = 0.033)	0.64 (0.033)	0.75 (0.025)	0.66 (0.045)	0.71 (0.038)		0.73 (0.273)
MKI	0.66 (0.031) (p = 0.011)	0.61 (0.042)	0.60 (0.036)	0.64 (0.045)	0.62 (0.047)		0.60 (0.039)
MYCN status	0.77 (0.038) (p = 0.001)	0.72 (0.048)	0.73 (0.052)	0.67 (0.031)	0.66 (0.053)		0.71 (0.043)