Table 6. Gene-gene interaction models for SNPs in HSCR risk by MDR analysis.

Number of factors	Best model*	Training accuracy	Testing accuracy	CVC	X2	P value	Odds Ratio (95%CI)
1	RET(rs7069590)	0.593	0.6337	10/10	33.705	<0.0001	2.22(1.69-2.91)
2	RET(rs3026737)-ARHGEF3(rs4681946)	0.5569	0.5743	10/10	11.830	0.0006	1.59(1.22-2.08)
3	RET(rs3026737)-RET(rs7069590)-ARHGEF3(rs11925835)	0.6193	0.6238	10/10	55.266	<0.0001	2.80(2.13-3.69)
4	RET(rs7069590)-ARHGEF3(rs11717604)-ARHGEF3(rs3732508)-CTNNAL1(rs7021366)	0.6477	0.5941	10/10	79.921	<0.0001	3.41(2.59-4.47)

*The best model was referred to as the one with the maximum testing accuracy and maximum cross-validation consistency (CVC); MDR = multifactor dimensionality reduction, CI = confidence interval, HSCR = Hirschsprung disease.