Design of a Clinical Decision Support System for Fracture Prediction Using Imbalanced Dataset

Table 3

Comparisons of predictive performance between models designed based on the methods proposed in this study and the one-sided selection methods [43] with the WDBC dataset [54].

Group	Accuracy	Sensitivity	Specificity	G-Mean	AUC
GA-SVM + Rand undersampling with OB1	97.64	97.64	97.64	0.9764	0.9915
GA-SVM + Rand undersampling with OB2	97.40	97.64	97.16	0.9740	0.9970
GA-SVM + cluster-based kNN undersampling with OB1	98.11	98.11	98.11	0.9811	0.9945
GA-SVM + cluster-based kNN undersampling with OB2	97.40	96.22	98.58	0.9739	0.9965
One-sided selection with OB1	97.84	97.97	97.64	0.9780	0.9956
One-sided selection with OB2	97.49	96.22	98.26	0.9723	0.9984