Fig. 2: Boxplots for performance comparison between different models/model structures in terms of testing set C-index. | npj Precision Oncology

Fig. 2: Boxplots for performance comparison between different models/model structures in terms of testing set C-index.

From: Autosurv: interpretable deep learning framework for cancer survival analysis incorporating clinical and multi-omics data

Fig. 2

Predictions measured on TCGA-BRCA and TCGA-OV datasets in three different cases: a mRNA + miRNA + clinical; b mRNA + clinical; c miRNA + clinical. AUTOSurv Entangle: AUTOSurv with “entangle” integration strategy; AUTOSurv Concat: AUTOSurv with “concatenate” integration strategy, more details about the alterations of AUTOSurv were illustrated in Supplementary Fig. 2; AUTOSurv Entangle No KL: AUTOSurv (with “entangle” integration strategy) without KL-annealing; Modif-SALMON: modified-SALMON. The p-value from two-sided Wilcoxon signed-rank test (null hypothesis \({{\rm{H}}}_{0}\): median difference is 0; versus alternative hypothesis \({{\rm{H}}}_{{\rm{A}}}\): median difference is not 0) is displayed between boxes.

Back to article page