Table 2.

Classification accuracy of attention-deficit/hyperactivity disorder (ADHD) vs. typically developing (TD) controls in models with different combinations of hyper-parameters (early stopping, batch size, and dropout rate).

Hyper-parameter	ATCACC	p-value
The original group with the completed dataset	0.8931 ± 0.0215
ES=100, BS=Batch, D=0.25	0.8859 ± 0.0142	0.415
ES=100, BS=Batch, D=0.2	0.8987 ± 0.0139	0.523
ES=100, BS=Batch, D=0.5	0.8954 ± 0.0164	0.801
ES=100, BS=8, D=0.25	0.8918 ± 0.0113	0.875
ES=10, BS=Batch, D=0.25	0.8885 ± 0.0118	0.585
ES=10, BS=Batch, D=0.5	0.899 ± 0.0121	0.484
ES=10, BS=8, D=0.25	0.8921 ± 0.0109	0.906
ES=10, BS=Stochastic, D=0.25	0.9046 ± 0.0159	0.216

Results presented as means ± standard deviations. Independent t-tests were used to compare the classification accuracy between the imputed and reference datasets. ATCACC, ADHD/TD classification accuracy; ES, earlystopping patience; BS, batch size; D, dropout rate.