Plot Kaplan-Meier Curves by Predicted Risk Group

Stratifies individuals into risk groups based on predicted risk scores from a fitted survival model, plots Kaplan-Meier survival curves per risk group, performs a log-rank test, and displays the concordance index (C-index) with confidence interval. The plot is saved as a PDF in "Results/".

Usage

plot_survival_performance(df_test, prediction, n_groups = 2, file_name = NULL)

Arguments

df_test

Data frame with the observed outcome of the test samples, in columns time and event (1 = event, 0 = censored), in the order of the predictions.

prediction

The output of compute_prediction() with task_type = "survival" (risk scores in $preds, C-index and its confidence interval in $c_index, $c_index_lower and $c_index_upper).

n_groups

Integer. Number of risk groups for stratification (default = 2, as in compute_prediction()). It must be at least 2 and not larger than the number of samples. Groups are labelled Low/High risk (2 groups), Low/Medium/High risk (3 groups) or Group 1 (lowest risk) to Group n (highest risk).

file_name

Optional character. The Kaplan-Meier plot is saved to "Results/Survival_KM_<file_name>.pdf" ("Results/Survival_KM.pdf" if NULL).

Value

Invisibly returns the ggsurvplot object for further customization. The plot (curves and number-at-risk table) is saved as a PDF.

Details

Risk groups are defined by quantiles of the predicted risk scores. Kaplan-Meier curves visualize survival per risk group, and a log-rank test assesses differences. The C-index and its 95% confidence interval are displayed in the plot subtitle.

When the predictions take few distinct values (e.g. tree-based models) and some quantiles are equal, samples with the same prediction are kept in the same group. Fewer groups than n_groups can then be formed (with a message); the function stops if all predictions are equal.

df_test must have one row per prediction, in the same order. Samples without a prediction (NA) are left out of the plot, with a message.

Examples

if (FALSE) { # \dontrun{
# res_survival: output of compute_features.training.ML(task_type = "survival")
pred <- compute_prediction(model = res_survival$Model,
                           test_data = X_test,
                           task_type = "survival",
                           time_var = time_test,
                           event_var = event_test)

plot_survival_performance(df_test = data.frame(time = time_test, event = event_test),
                          prediction = pred,
                          n_groups = 3,
                          file_name = "Example")
} # }