Table 4. Open in new tab The...

Model:	RF	GBM	SVM
Accuracy	69.5 ± 2	61.8 ± 2	70.8 ± 3	61.8 ± 1	71.9 ± 2	62.9 ± 3
F1-score	67.8 ± 2	59.1 ± 2	70.0 ± 2	60.7 ± 2	70.1 ± 2	60.6 ± 2
AUPRC	75.9 ± 1	66.6 ± 4	77.8 ± 1	65.2 ± 3	79.2 ± 1	67.7 ± 3
MCC	0.392	0.239	0.417	0.235	0.440	0.261
AUROC	75.2 ± 2	66.3 ± 2	77.3 ± 2	66.2 ± 2	78.1 ± 2	67.9 ± 3

Model:	RF	GBM	SVM
Accuracy	69.5 ± 2	61.8 ± 2	70.8 ± 3	61.8 ± 1	71.9 ± 2	62.9 ± 3
F1-score	67.8 ± 2	59.1 ± 2	70.0 ± 2	60.7 ± 2	70.1 ± 2	60.6 ± 2
AUPRC	75.9 ± 1	66.6 ± 4	77.8 ± 1	65.2 ± 3	79.2 ± 1	67.7 ± 3
MCC	0.392	0.239	0.417	0.235	0.440	0.261
AUROC	75.2 ± 2	66.3 ± 2	77.3 ± 2	66.2 ± 2	78.1 ± 2	67.9 ± 3

Table 4.

The middle-exclusion filter effect on performance metrics. Three classical models were each trained and tested with and without the middle exclusion filter applied to the full dataset. The results reflect two rounds of 5-fold cross-validation on 90% of the data (10% withheld for testing but not used here)

Model:	RF		GBM		SVM
Filter:	Middle exclusion	None	Middle exclusion	None	Middle exclusion	None
Accuracy	69.5 ± 2	61.8 ± 2	70.8 ± 3	61.8 ± 1	71.9 ± 2	62.9 ± 3
F1-score	67.8 ± 2	59.1 ± 2	70.0 ± 2	60.7 ± 2	70.1 ± 2	60.6 ± 2
AUPRC	75.9 ± 1	66.6 ± 4	77.8 ± 1	65.2 ± 3	79.2 ± 1	67.7 ± 3
MCC	0.392	0.239	0.417	0.235	0.440	0.261
AUROC	75.2 ± 2	66.3 ± 2	77.3 ± 2	66.2 ± 2	78.1 ± 2	67.9 ± 3

Model:	RF		GBM		SVM
Filter:	Middle exclusion	None	Middle exclusion	None	Middle exclusion	None
Accuracy	69.5 ± 2	61.8 ± 2	70.8 ± 3	61.8 ± 1	71.9 ± 2	62.9 ± 3
F1-score	67.8 ± 2	59.1 ± 2	70.0 ± 2	60.7 ± 2	70.1 ± 2	60.6 ± 2
AUPRC	75.9 ± 1	66.6 ± 4	77.8 ± 1	65.2 ± 3	79.2 ± 1	67.7 ± 3
MCC	0.392	0.239	0.417	0.235	0.440	0.261
AUROC	75.2 ± 2	66.3 ± 2	77.3 ± 2	66.2 ± 2	78.1 ± 2	67.9 ± 3

This Feature Is Available To Subscribers Only