Table 5. MDR analysis for the risk of prostate cancer prediction in an Chinese population.

Best interaction models	Cross-validation	Average prediction error	P-valuea
rs1468033	100/100	0.4566	0.0001
rs2295080 rs1468033	100/100	0.3451	p < 0.0001
rs17036508 rs2295080 rs1468033	100/100	0.3434	p < 0.0001
age rs17036508 rs2295080 rs1468033	99/100	0.4066	p < 0.0001
age rs17036508 rs2295080 rs1468033 rs7250897	78/100	0.4254	p < 0.0001
BMI smoking_status race rs17036508 rs2295080 rs1468033	45/100	0.4467	p < 0.0001
smoking_status age race rs17036508 rs2295080 rs1468033 rs7250897	61/100	0.4022	p < 0.0001
BMI smoking_status age race rs17036508 rs2295080 rs1468033 rs7250897	100/100	0.5066	p < 0.0001

MDR: multifactor dimensionality reduction.

The best model with maximum cross-validation consistency and minimum prediction error rate was in bold.

^aP-value for 1000-fold permutation test.