Identifying the Gene Signatures from Gene-Pathway Bipartite Network Guarantees the Robust Model Performance on Predicting the Cancer Prognosis

Table 2

The results of predicting the reoperative treatment response of breast cancer in original and swap analyses.

		Our model				MAQC-II candidate model
		SP	SE	ACC	MCC	SP	SE	ACC	MCC
Original analysis	Training	0.928	0.455	0.808	0.444	0.847	0.569	0.775	0.433
	Validation	0.882	0.467	0.820	0.332	0.729	0.667	0.720	0.301
Swap analysis	Training	0.988	0.200	0.870	0.343	0.899	0.522	0.837	0.454
	Validation	1.000	0.152	0.785	0.343	0.959	0.212	0.769	0.267

In the prediction, pCR was defined as positive sample.
SP, SE, ACC, and MCC represented specificity, sensitivity, accuracy, and Matthew’s correlation coefficient, respectively.