Figure 4From: A semi-parametric Bayesian model for unsupervised differential co-expression analysisDCIM derived contexts in breast cancer data are predictive of patient survival. DCIM was used to identify the top two contexts marked (1) and (2) in Figure 3 and survival analysis was performed. A) Kaplan-Meier curves for the two contexts. B) Clinical, molecular, and computational parameters and their pairwise combinations were used to fit one-parameter and two-parameter Cox regression and the model fit was assessed using the Akaike Information Criterion (AIC). The model combining DCIM contexts and AURKA expression defined patient groups best predicts patient survival (red box) http://eh3.uc.edu/gimm/dcim.Back to article page