Table 2.

Performances of different hyper-parameter values of TCGA three-omics data.

	Epoch	Additional Hidden Layer Shape	Bottleneck Layer Shape	Normalization	Experiment	Survival-Related Node Number	3-Omics C-Index
No. of hyper-parameters used	5	500	200	Standard Normalization	DL	53	0.684 (SE 0.023)
No. of bottleneck nodes	10	500	100	Standard Normalization	DL	24	0.656 (SE 0.022)
	10	500	200	Standard Normalization	DL	59	0.668 (SE 0.023)
	10	500	300	Standard Normalization	DL	124	0.668 (SE 0.02)
	10	1000	100	Standard Normalization	DL	26	0.673 (SE 0.027)
	10	1000	300	Standard Normalization	DL	118	0.683 (SE 0.02)
	10	1000	500	Standard Normalization	DL	207	0.678 (SE 0.02)
No. of epochs	1	500	200	Standard Normalization	DL	70	0.677 (SE 0.02)
	5	500	200	Standard Normalization	DL	53	0.684 (SE 0.023)
	15	500	200	Standard Normalization	DL	83	0.676 (SE 0.026)
	30	500	200	Standard Normalization	DL	95	0.661 (SE 0.023)
	50	500	200	Standard Normalization	DL	124	0.663 (SE 0.027)
	1	1000	300	Standard Normalization	DL	115	0.679 (SE 0.02)
	5	1000	300	Standard Normalization	DL	103	0.672 (SE 0.02)
	15	1000	300	Standard Normalization	DL	120	0.666 (SE 0.023)
	30	1000	300	Standard Normalization	DL	162	0.659 (SE 0.023)
	50	1000	300	Standard Normalization	DL	188	0.67 (SE 0.023)
Hidden layers	5	1000, 500	100	Standard Normalization	DL	24	0.68 (SE 0.025)
	5	1000, 500	200	Standard Normalization	DL	86	0.668 (SE 0.023)
	5	1000, 600	200	Standard Normalization	DL	87	0.661 (SE 0.023)

DL, deep learning; SE, standard error.