Fig. 5: Validation of EMIT on independent inhouse lung cancer dataset.
From: Development of a deep learning model for cancer diagnosis by inspecting cell-free DNA end-motifs
a ROC curve of linear projection of representations extracted from EMIT in the diagnosis of cancer. b, c End-motifs exhibited differential attribution scores. The boxplot represents the distribution of the attribution score, where the box spans the interquartile range (IQR) from the first quartile (Q1) to the third quartile (Q3). The median is denoted by the center line within the box, while the whiskers extend to encompass data points within 1.5 times the IQR from Q1 and Q3. d, e End-motif attention netoworks exhibited differential topology. *** represented adjusted p-value < 0.05.
