Combination of Genetic Markers and Age Effectively Facilitates the Identification of People with High Risk of Preeclampsia in the Han Chinese Population

Table 3

Interaction among multiple polymorphic loci.

Model^1	Training Bal. Acc	Testing Bal. Acc.	CV consistency	Sign test (P)
rs1799724	0.5639	0.5645	10/10	7 (0.1719)
rs1800896 and rs1799724	0.5978	0.5592	9/10	8 (0.0547)
rs1800896, rs1800629, and rs1799724	0.6163	0.5196	4/10	6 (0.3770)
rs2549782, rs1695, rs1800896, and rs1799724	0.6391	0.5052	3/10	6 (0.3770)
rs2549782, rs1695, rs1800896, rs1799724, and rs4762	0.6714	0.5362	8/10	7 (0.1719)
rs2549782, rs1695, rs1800896, rs1800629, rs1799724, and rs4762	0.7030	0.5691	8/10	8 (0.0547)
rs2549782, rs1695, rs1800896, rs1800629, rs1799724, rs4762, and rs7412	0.7294	0.5853	10/10	10 (0.0010)
rs2549782, rs1695, rs2070744, rs1800896, rs1800629, rs1799724, rs4762, and rs7412	0.7378	0.5633	10/10	8 (0.0547)

Only the best interaction models with no larger than eight features are shown. The significant and best model is shown in bold.
“Training Bal. Acc,” training balanced accuracy; “Testing Bal. Acc,” testing balanced accuracy; “CV consistency,” 10-fold cross-validation consistency.