Integrated survival metrics

Integrated survival metrics summarize model performance across multiple evaluation time points into a single value by integrating dynamic survival metrics over time.

Input requirements

• truth: a survival::Surv() object

• ...: list column of data frames containing .eval_time, .pred_survival, and .weight_censored columns

Available metrics

brier_survival_integrated()

Direction: minimize. Range: [0, 1]

See also

dynamic-survival-metrics for time-dependent survival metrics

static-survival-metrics for static survival metrics

linear-pred-survival-metrics for linear predictor survival metrics

vignette("metric-types") for an overview of all metric types

Examples

library(dplyr)
data("lung_surv")

head(lung_surv)
#> # A tibble: 6 × 4
#>   .pred            .pred_time surv_obj .pred_linear_pred
#>   <list>                <dbl>   <Surv>             <dbl>
#> 1 <tibble [5 × 5]>       324.     306               5.78
#> 2 <tibble [5 × 5]>       476.     455               6.17
#> 3 <tibble [5 × 5]>       521.    1010+              6.26
#> 4 <tibble [5 × 5]>       368.     210               5.91
#> 5 <tibble [5 × 5]>       506.     883               6.23
#> 6 <tibble [5 × 5]>       324.    1022+              5.78

lung_surv |>
  brier_survival_integrated(truth = surv_obj, .pred)
#> # A tibble: 1 × 3
#>   .metric                   .estimator .estimate
#>   <chr>                     <chr>          <dbl>
#> 1 brier_survival_integrated standard       0.158