Skip to main content
Fig. 2 | BMC Medical Research Methodology

Fig. 2

From: Survival prediction models since liver transplantation - comparisons between Cox models and machine learning techniques

Fig. 2

a Predicted survival probabilities for 3 new hypothetical patients using the Cox model with all variables (solid lines), the tuned RSF (short dashed lines) and the tuned NN with 1 hidden layer (long dashed lines). The green lines correspond to a reference patient with the median values for the continuous and the mode value for categorical variables. The patient in the orange line has diabetes (the other covariates as in reference patient). The patient in the red line has been transplanted before and has diabetes simultaneously (the other covariates as in reference patient). Values for 10 prognostic variables for the reference patient are provided in Table 2 of Additional file 1. b Predicted survival probabilities for 3 patients selected from the test data based on the Cox model with all variables (solid lines), the tuned RSF (short dashed lines) and the tuned NN with 1 hidden layer (long dashed lines). Green lines correspond to a patient censored at 1.12 years. Patient in the orange line was censored at 6.86 years. Patient in the red line died at 0.12 years. Values for 10 prognostic variables for the patients are provided in Tables 3-5 of Additional file 1

Back to article page