Table 2.

Mean AUC ± one standard deviation for 10-fold cross validation testing different architectures and data combinations.

	0 Hidden layers	1 Hidden layer	2 Hidden layers		3 Hidden layers
	1	12-1	8-8-1	12-12-1	12-8-4-1	12-12-4-1
PLCO	0.59 ± 0.078	0.65 ± 0.033	0.64 ± 0.040	0.66 ± 0.027	0.65 ± 0.037	0.65 ± 0.037
NHIS-1	0.58 ± 0.129	0.59 ± 0.131	0.60 ± 0.132	0.57 ± 0.129	0.56 ± 0.130	0.60 ± 0.131
NHIS-2	0.59 ± 0.109	0.60 ± 0.109	0.62 ± 0.109	0.57 ± 0.107	0.55 ± 0.108	0.60 ± 0.109
NHIS-3	0.57 ± 0.096	0.59 ± 0.096	0.58 ± 0.096	0.57 ± 0.096	0.56 ± 0.095	0.58 ± 0.096
NHIS-4	0.59 ± 0.088	0.61 ± 0.089	0.63 ± 0.088	0.59 ± 0.089	0.58 ± 0.089	0.61 ± 0.088
NHIS-5	0.60 ± 0.084	0.61 ± 0.084	0.61 ± 0.084	0.60 ± 0.084	0.57 ± 0.084	0.58 ± 0.084
NHIS-6	0.61 ± 0.080	0.62 ± 0.080	0.61 ± 0.080	0.60 ± 0.080	0.59 ± 0.080	0.60 ± 0.080
NHIS-7	0.59 ± 0.077	0.60 ± 0.077	0.60 ± 0.077	0.59 ± 0.077	0.57 ± 0.076	0.59 ± 0.077
NHIS-8	0.63 ± 0.075	0.61 ± 0.075	0.64 ± 0.075	0.62 ± 0.075	0.61 ± 0.075	0.62 ± 0.075
PLCO+NHIS-1	0.72 ± 0.051	0.80 ± 0.064	0.80 ± 0.064	0.80 ± 0.064	0.80 ± 0.064	0.80 ± 0.064
PLCO+NHIS-2	0.66 ± 0.054	0.76 ± 0.064	0.76 ± 0.063	0.75 ± 0.064	0.76 ± 0.064	0.75 ± 0.064
PLCO+NHIS-3	0.64 ± 0.053	0.72 ± 0.062	0.74 ± 0.061	0.72 ± 0.062	0.73 ± 0.062	0.72 ± 0.062
PLCO+NHIS-4	0.61 ± 0.054	0.71 ± 0.060	0.72 ± 0.060	0.71 ± 0.061	0.72 ± 0.060	0.72 ± 0.060
PLCO+NHIS-5	0.58 ± 0.054	0.71 ± 0.059	0.71 ± 0.059	0.71 ± 0.059	0.70 ± 0.059	0.70 ± 0.059
PLCO+NHIS-6	0.57 ± 0.053	0.69 ± 0.058	0.70 ± 0.058	0.69 ± 0.058	0.70 ± 0.058	0.69 ± 0.058
PLCO+NHIS-7	0.56 ± 0.052	0.68 ± 0.057	0.68 ± 0.057	0.68 ± 0.057	0.68 ± 0.057	0.68 ± 0.057
PLCO+NHIS-8	0.54 ± 0.053	0.69 ± 0.056	0.70 ± 0.056	0.69 ± 0.056	0.69 ± 0.057	0.69 ± 0.055

In the leftmost column, the NHIS-X stands for the NHIS data with a cutoff year of X for ovarian cancer diagnosis prior to the survey. The 2nd row headers give the number of neurons in the hidden layers and output layer. The bold indicates the selected model for further study.