Figure 4
Predictive values for CLEAR Score model versus prior models. The predictive value is represented by the likelihood ratio test and adequacy index. The adequacy index is a measure of the evaluated model (as a subset of the full model), represented as a % relative to the full model, which does include both subtype and CLEAR Score model). (A) Predictive value of CCA/B subtyping (110 transcripts) (Gulati et al.) and CLEAR score model for cancer-specific survival (CSS) using TCGA dataset. (B) Predictive value of CCA/B subtyping (34 transcripts) (Brooks et al.) and CLEAR score model for CSS using TCGA dataset. (C) Predictive value of sunitinib-responsive subtyping (Beuselinck et al.) and CLEAR Score model for progression free survival (PFS) using E-MTAB-3267 datasets. These three comparison result showed the addition of the CLEAR score model to one containing the CCA/B subtype model significantly improves the predictive value of the final model, but there is no significant difference when the subtype model is added to the CLEAR model.
