Jump to content

Survival Kit, software

From Wikiversity
Subject classification: this is a statistics resource.
Educational level: this is a tertiary (university) resource.

This resource is to describe the usage of Survival Kit[1][2][3] software package used for survival analysis (the paper). The focus will be on step by step description how to run the basic examples with fixed effects. The real advantage of the Survival Kit is its ability to handle very large databases (up to several million records) including random effects. This is however outside of scope of this beginner guide, while novice users should get familiar with the basic principles first.
The description deals with Survival Kit v6.06 available online at boku.ac.at.

Release notes
  • 15. March 2010 - Wikiversity page has been created to assist Survival Kit v3.12 users.
  • 02. November 2010 - Description on the page updated for SK v6.06, the program was uploaded to its non-official storage space.
  • 08. November 2010 - The executable files updated on the [Mediafire storage space], but the link remained the same. Upload to the official BOKU site still pending.
  • 23. November 2010 - The Survival Kit v6.06 is now reachable on its official website. See the World congress paper for more details. The temporary storage place was canceled.


General description of the Survival Kit

[edit | edit source]

The Survival Kit is intended for survival analysis using proportional hazards model with a single response time. These models describe the hazard function of each individual (i.e. its limiting probability of dying at time t, given it is still alive just prior to t) as the product of a baseline hazard function and a positive (exponential) function of explanatory covariates. The baseline hazard function can be parametric (Weibull model) or left unspecified (Cox model[4]).
When the failure time variable is discrete (i.e. has very few values) and many observations with the same failure times, many ties occur between failure times. In such case, the Cox model is not valid any more, but the "grouped data" approach[5] should be used.

The Survival Kit consists from 3 main programs:

  • Prepare - for database preparation, internal recoding
  • Cox - for non parametric Cox model[4]
  • Weibull - for grouped data[5] or parametric Weibull models

Prerequisites

[edit | edit source]

The reader should be familiar with the basic concepts of survival analysis. Basic knowledge of computer usage is expected.

Users with practical experiences are encouraged to share them here. If something is not clear ask here or here.

See also

[edit | edit source]
Search for Survival analysis on Wikipedia.

References

[edit | edit source]
  1. Ducrocq, V., Sölkner J.(1998) Proc. 6th World Cong. on Genet. Appl. To Livest. Prod., 27: 447-448.
  2. Sölkner J., Ducrocq, V., (1999) Workshop on Genetic Improvement of Functional Traits in Cattle - Longevity, Jouy-en-Josas, France
  3. Ducrocq, V., Sölkner, J. and Mészáros G. (2010). Survival Kit v6 – a Software Package for Survival Analysis. In: 9th World Cong. Genet. Appl. Livest. Prod., Leipzig, Germany.
  4. 4.0 4.1 Cox, D.R. (1972) J. R. Stat. Soc., Series B, 34: 187-220.
  5. 5.0 5.1 Prentice, R. and Gloeckler, L. (1978) Biometrics, 34:57-67.