Censored Data Analysis - Exercise class

Back to /~vavraj/

• SIS page: NMST511
• Schedule: officially on Thursday 14:00 - 15:30 in classroom K5
• Language: materials in English, instructed in Czech (unless anybody requires English)
• Instructor e-mail: vavraj@karlin.mff.cuni.cz
• Lecture web page (from previous years): http://www.karlin.mff.cuni.cz/~kulich/vyuka/cens/index.html
• Note: Many exercises will be replaced by lectures untill sufficient theoretical background is covered.
  • Hence, exercises will be scattered irregularly within the term (mostly at its end).
  • Course credit will be given for completing given homework assignments in pre-specified deadlines (usually a morning before the next exercise).
  • Exercise class will be opened with discussion over previous homework.
  • We will go through the materials for exercise class together.
  • There will be time for you to ask questions, work on given assignments, …
  • No attendance requirement for course credit.

Credit

• Satisfactory solution to given assignments by the prescribed deadline.
  (If your solution will not be satisfactory enough, you might be asked for revision.)
  Send your solution via email to address vavraj@karlin.mff.cuni.cz.
  Recommended form: Rmarkdown output (PDF or HTML)

Overview

Exercise 1 (16 October)

• Topic: Parametric regression models for censored data
• Dataset: pbc dataset from library(survival)
• R-scripts: ignoring_censoring.R
• Assignment: Find suitable model for survival time and interpret it.
• Deadline: Thursday 23 October 9:00

Exercise 2 (20 November)

• Topic: Non-parametric estimation of cumulative hazard and survival function
• Dataset: fans, km_all.RData, mort.m and mort.f from mort.RData
• Assignment: Calculate and plot [NA], [KM], [FH] estimates both manually and using survfit function from library(survival). Perform short (or long if you wish) simulation study fo Hall-Werner confidence bands. Compare survival estimators with the output from ČSÚ: males and females.
• Deadline: Thursday 27 November 9:00

Exercise 3 (11 December)

• Topic: Testing equality of censored distributions
• Datasets: km_all.RData, data(veteran)
• Assignment: Calculate and plot [KM], [NA] estimates and smoothed estimator of hazard function when differentiating different groups. Perform two-sample tests and decide (based on the plots) which test statistic would be the most appropriate. Perform several real data analyses and simulation study.
• Deadline: Thursday 18 December 9:00

Exercise 4 (18 December)

• Topic: Building Cox models for censored data (constant covariates)
• Dataset: pbc dataset from library(survival)
• Assignment: Perform exploratory analysis focused on the influence of covariates on the survival probability. Build a reasonable Cox model (starting from simple univariate models). Compare your final Cox model to your final model from Exercise 1.
• Deadline: no report required

Exercise 5 (8 January)

• Topic: Time-varying covariates in the Cox model
• Dataset: jasa and heart dataset from library(survival)
• Assignment: Try to reproduce heart dataset from jasa. Build and interpret Cox model with covariate indicating the time of heart transplantation and other fixed covariates. Does transplantation help patients to survive longer? Plot estimated survival functions.
• Deadline: no report required

Exercise 6 (18 December + 8 January)

• Topic: Cox model - homework assignment
• Dataset: aids dataset from library(JM)
• Assignment: Perform exploratory analysis focused on the influence of covariates on the survival probability. Take into account the longitudinal aspect of the data. Build a reasonable Cox model (starting from simple univariate models).
• Deadline: end of exam period, Friday 13 February

Bonus Exercise (not scheduled)

• Topic: Generalizations of the Cox model
• Dataset: pbc and cgd datasets from library(survival)
• Assignment: None, just admire what else could be done with the Cox model.