Table 5.

SNP–SNP interaction models of the MIR3142HG gene analyzed by the GMDR^a.

Model	Training Bal. Acc	Testing Bal. Acc.b	CVC
rs2961920	0.5391	0.5159	6/10
rs2431689, rs58747524	0.5487	0.5145	7/10
rs2431689, rs2961920, rs58747524	0.5568	0.5034	5/10
rs1582417,rs7727115, rs2961920, rs58747524	0.5660	0.4880	4/10
rs1582417, rs7727115, rs17057846, rs2961920, rs58747524	0.5754	0.5036	5/10
rs1582417, rs2431689, rs7727115, rs17057846, rs2961920, rs58747524	0.5802	0.5149	10/10

a Whole dataset statistics: Training Balanced Accuracy, 0.5782; Training Accuracy, 0.5782; Training Sensitivity, 0.5302; Training Specificity, 0.6261; Training Odds Ratio, 1.8899 (1.3227, 2.7003); Training χ2 (P), 12.3300 (p = 0.0004); Training Precision, 0.5860; Training Kappa, 0.1563; Training F-Measure, 0.5567.

GMDR: Generalized Multifactor dimensionality reduction; Bal. Acc: Balanced accuracy; CVC: Cross‐validation consistency.