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

Table 3

The results of predicting the overall survival milestone outcome of acute myeloma leukemia in original and swap analyses.

		Our model					86-probe-set model
		SP	SE	ACC	MCC	AUC	SP	SE	ACC	MCC	AUC
Original analysis	Training	0.697	0.837	0.776	0.542	0.776	0.758	0.733	0.743	0.486	0.746
	Validation	0.574	0.600	0.584	0.170	0.587	0.362	0.800	0.532	0.172	0.581
Swap analysis	Training	0.830	0.700	0.779	0.533	0.765	1.000	0.433	0.779	0.564	0.717
	Validation	0.545	0.756	0.664	0.308	0.655	0.712	0.523	0.605	0.236	0.618

In the prediction, high-risk patient was defined as positive sample.
AUC represented the area under the ROC curve.
See notes under Table 2 for more information.