Table 2.

Comparison of recurrence prediction results of ensemble learning models using different feature representations.

Feature representation	Acc	Recall	Prec	F1 score
Personal and	0.6062	0.6164	0.6019	0.6039
clinical indicators	± 0.0877	± 0.1429	± 0.0930	± 0.1027
AP	0.7224	0.6946	0.7528	0.7156
	± 0.0834	± 0.0944	± 0.1231	± 0.0804
PVP	0.6438	0.6782	0.6409	0.6570
	± 0.1117	± 0.1122	± 0.1019	± 0.1014
DP	0.6343	0.6909	0.6331	0.6560
	± 0.0690	± 0.0413	± 0.0872	± 0.0520
AP+	0.7495	0.7673	0.7402	0.7502
other indicators	± 0.0629	± 0.1051	± 0.0525	± 0.0710
PVP+	0.6824	0.6927	0.6844	0.6846
other indicators	± 0.0783	± 0.0941	± 0.0780	± 0.0771
DP+	0.6819	0.6309	0.7168	0.6630
other indicators	± 0.0659	± 0.0912	± 0.1273	± 0.0756

Each result is represented by the mean of 5 experiments and 95% CI.