Pooling breast cancer datasets has a synergetic effect on classification performance and improves signature stability.

Both the DLCV error as well as the error on the validation set show a highly significant correlation with the number of artificial datasets that is pooled, Pearson correlation of -0.91 ( p = 1.1 e - 24) and -0.95 ( p = 1.4 e - 31), respectively.

also in 82,756 other papers

These results indicate a positive trend in the number of datasets pooled and the number of genes selected in the signatures (Pearson correlation of 0.38 with p = 1.9 e - 3).

also in 6,846 other papers

There is a clear trend indicating that the DLCV error decreases when more datasets are pooled (see Figure 6A ).

also in 12,254 other papers