Skip to main content

Table 2 AUC classification performance for different classification tasks

From: On Predicting lung cancer subtypes using ‘omic’ data from tumor and tumor-adjacent histologically-normal tissue

Classification Task Omic Feature selection with ReliefF Feature selection with Limma
AUC 95 % C.I. BSS AUC 95 % C.I. BSS
TAHNADC vs. TumorADC G 0.99 0.97–1.0 0.89 0.94 0.82–1.0 0.73
M 1.0 1.0–1.0 0.99 0.81 0.58–0.97 0.17
TAHNSCC vs. TumorSCC M 1.0 0.99–1.0 0.94 0.99 0.96–1.0 0.66
TumorADC vs. TumorSCC G 0.89 0.83–0.96 0.29 0.90 0.89–0.9 0.81
M 0.97 0.94–0.99 0.71 0.89 0.74–1.0 0.38
TAHNADC vs. TAHNSCC M 1.0 1.0–1.0 0.92 1.0 1.0–1.0 0.99
TAHN-TumorADC vs. TAHN-TumorSCC M 0.92 0.89–0.95 0.42 0.94 0.87–1.0 0.56
  1. G: gene expression, M: DNA methylation. The Brier Skill Score is a measurement of calibration of the classifier. A positive value on the BSS means that the classifier is well calibrated. A baseline classification is the work by Chang and Ramoni [22] which obtained an accuracy of 0.95 in the classification task TumorADC vs. TumorSCC