Graph based fusion of miRNA and mRNA expression data improves clinical outcome prediction in prostate cancer

BMC Bioinformatics

Table 1 Comparison of Boosting Results.

	M	IPEC (median)	IQR	p-value
only miRNA	98	5.90	0.88	< 0.001
only mRNA	100	5.82	0.87	< 0.001
both no graph	99	5.79	0.86	< 0.001
both with graph	99	5.46	1.20	-

The table shows the number of boosting steps M for every CoxBoost model and the IPEC (median and IQR) of 500 bootstrap runs. The number of boosting steps were determined using the whole data set. Lower IPEC scores indicate better prediction accuracy. The p-value is the result of a one-sided Wilcoxon test (unpaired) comparing the single data set prediction models and the prediction model without graph with the combination incorporating the bipartite graph.

ISSN: 1471-2105