Skip to main content
. Author manuscript; available in PMC: 2016 Aug 29.
Published in final edited form as: Conf Proc IEEE Eng Med Biol Soc. 2009;2009:4162–4165. doi: 10.1109/IEMBS.2009.5333937

Figure 4.

Figure 4

Area under the curve (AUC) plots representing biomarker detection efficiency for several feature ranking metrics. A larger AUC indicates higher detection efficiency. The optimal ranking metric, selected using maximum likelihood estimation (MLE), is more efficient compared to significance analysis of microarrays (SAM), a standard ranking method. The use of sub-optimal knowledge (sub-opt) when selecting the ranking metric decreases detection efficiency. When using randomly selected genes as knowledge, detection efficiency is random (control).