This course covers both the theoretical and practical aspects of the methodology relating to survival data.
Benefits to participants
Participants will learn how to organize the survival data base, how to choose the appropriate unit of time, when to use discrete methods and when to use continuous methods, how to deal with censored data (left, right, interval), how to choose the correct SAS procedure, what to do with the proportionality assumption, how to calculate an R squared statistic and residuals, when and how to correct for unobserved heterogeneity, what the frequency is for measuring independent variables, what to do if there is more than one type of event, how to test the assumption concerning non-informative censor, and how to know whether a model fits well to the data.
This course is of benefit to employees required to analyse data relating to events that span a period of time (survival data, event analysis data, reliability data).
- Survival Data
- Choice of time axis
- Censoring (left, right, interval)
- Various parametric Models (exponentiel, Weibull, Gompertz, logistic, etc)
- Kaplan-Meyer estimator
- Interpretation of parameters
- Proportional risk models
- AFT models
- Partial likelihood
- Competing risks
- Time-dependent variables
- Analysis of discrete data
- Analysis of sensitivity to censor data
- Choice of models and quality of data adjustment
- Verifying the proportionality assumption
- Heterogeneity and time dependency
- R squared and standardized coefficient
- Repetitive events
Some experience with and knowledge of the principles of multiple regression and of the basic concepts of statistical inference. A basic knowledge of SAS would be useful (familiarity with PROC REG and how to create/modify simples files with the DATA step)
Duration: 4 days