TABLE 1.

Misclassified training samples when using various classification models and omics data.

Color coding indicates the number of times any of the 40 DCIS cases were misclassified as IBC in the training data; green, once; brown, twice; red, three times. CpG methylation data with the RF model was not used in these analyses since it misclassified a total of 34 DCIS cases, which was considered too many. We considered for further investigation only those DCIS cases in the training cohort that were repeatedly misclassified by these model–data combinations more than once.