Skip to main content
Fig. 4 | BioData Mining

Fig. 4

From: Integration and comparison of different genomic data for outcome prediction in cancer

Fig. 4

Agreement in the prognostic prediction by cancer type and data type. The figure shows the Cohen’s Kappa agreement of the risk assessment based on the median of the prognostic index generated by each model. Each heatmap shows the comparison of the models generated between data types and algorithms. Cells shown in squares correspond to the comparisons between the models of the three algorithms (CPSO, NFS, LASSO) for the same data type. The event proportion in each cancer type is shown in parenthesis. Within heatmaps, blue colors denote lower kappa value, white denotes intermediate values, and red denotes high kappa values. For comparison, the scatter-plots shown aside the color-coding corresponds to examples of prognostic indexes pairs having 0, 0.25, 0.5, and 1 of kappa values. MUT data did not generate risk groups in OV and GBM and were omitted. EXPR: Gene Expression. MIRNA: micro RNAs. CNA: Copy Number Alteration. MUT: Somatic mutations. LASSO (Least Absolute Shrinkage and Selection Operator). NFS: Network Feature Selector. CPSO: Constrained Particle Swarm Optimization. BRCA: Breast Cancer. LUAD: Lung Adenocarcinoma. OV: Ovary Cystadenocarcinoma. GMB: Glioblastoma Multiform

Back to article page