0

Is it possible to do bootstrap in survival analysis, says Kaplan-meier estimates?

If so, how should I do it?

I have read some articles, but they are too complicated. There's a function called BootKM in R, can anyone briefly explain how it works?

Siguza
  • 21,155
  • 6
  • 52
  • 89
Sylvia Liu
  • 1
  • 2
  • Here's the code for book, it does what the answer you got suggests: https://rdrr.io/cran/Hmisc/src/R/bootkm.s – kqr Jul 05 '22 at 16:11

1 Answers1

1

Yes, you can use bootstrapping w/ survival analysis and Kaplan-Meiers.

I don't know the inner workings of BootKM, but bootstrapping in survival analysis works the same as any other kind of bootstrapping 1. Generate n replicates of your patient-level data using resampling. 2. Calculate the statistics of interest based on the replicates. 3. 2.5th/97.5th percentiles of statistic effectively represent confidence interval on statistic of interest.

Keep in mind that when resampling your survival data, the maximum follow-up of the curve will change and your ability to generate confidence intervals only goes so far as the shortest of those curves.

jrdnmdhl
  • 1,935
  • 18
  • 26