19 Apr module 17: survival analysis
- module 17: survival analysis
17/09/2020 - 18/09/2020
9:30 am - 5:00 pm
Thu 17/09: 9:30 am - 5:00 pm | online
Fri 18/09: 9:30 am - 5:00 pm | online
week 2 | register now
Time-to-event data are abundant in many fields: time to death or relapse in a medical setting, time to graduation or recidivism in social sciences or time to breakdown of machine parts in an industrial setting. Clearly, the time to an event can only be observed for those who already experienced the event of interest: your refrigerator broke down after 5.2 years, while your oven is still operational after 8 years. For the latter, we only know the time to breakdown is more than 8 years. Where standard regression techniques mishandle the information from these censored observations, Survival analysis efficiently uses all available information in a correct way. Day 1 introduces the basic concepts of Survival analysis, illustrated with hands on computer labs in R. Day 2 discusses the specificities of sample size calculations for time-to-event data, with standard approaches in R and SAS, more flexible approaches in STATA and an introduction to building simulations to test the impact of various deviations from the basic assumptions. Computer labs will be in R only.
Day 1: Introduction to survival analysis
Day 2: Power and sample size calculations for time-to-event trials
To take this course, you should have a good working knowledge of the principles and practice of multiple linear regression, as well as elementary statistical inference. Familiarity with R and Rstudio is recommended. !!! Bring your own laptop, with R (and Rstudio) installed !!!
background readings: Modelling Survival Data in Medical Research (2015). David Collett. Third edition, Chapman and Hall/CRC. ISBN 978-1-439-85678-9
Dries Reynders is statistical consultant for StatGent-Crescendo at UGent. He focusses on survival analysis in clinical trials and observational data.