Skip to main content
Fig. 2 | BMC Medical Genomics

Fig. 2

From: Interpretable deep neural network for cancer survival analysis by integrating genomic and clinical data

Fig. 2

Graphical visualization of the node values in the second hidden layer (H2) and clinical layer. a Heatmap of the 31 nodes (i.e., thirty H2 nodes and one clinical node). The horizontal dashed line in red distinguishes two risk groups, where the upper/lower partition belongs to high risk/low risk patients. The top dot plot indicates the nodes’ significance. A logrank test was conducted for each node within two risk groups in the scale of -log10(p-values), where red indicates statistical significance, and blue shows insignificance. The plot in the right panel displays the prognostic index (PI) with each corresponding sample. b–c Kaplan-Meier plots of the top two nodes

Back to article page