1998 SSC Annual Meeting
1998 Annual Meeting of the SSC in SherbrookeBIOSTATISTICS WORKSHOP
Modern Methods in Survival Analysis
May 31, 1998
The theory of counting processes and martingales has had far-reaching effects on methods for analyzing time to event data. In the last 10 years this approach has not only provided methodologic breakthroughs for establishing the asymptotic distribution theory and operating characteristics of most nonparametric tests and estimators, but it has also provided an intuitive approach to new statistical methods, and, somewhat surprisingly, the design of software for the analysis of censored data.
This one day course will provide an introduction to martingale methods in survival analysis, with an emphasis on its intuitive basis and on examples. The material presented will include: A brief survey of the relevant notation and results from counting processes, including martingale representations of censored data statistics, stochastic integration with respect to martingales, variance formulas that use the concept of predictable quadratic variation, and martingale central limit theorems.
Applications to proportional hazards regression, including the use of PROC PHREG in SAS and the COXPH function in S-Plus for model fitting and regression diagnostics.
Applications to newer methods being developed for correlated and repeated event survival data, including the Andersen-Gill extension of the Cox model, marginal models for correlated failure time data and frailty models.
The workshop will be given by T.R. Fleming (University of Seattle), D.P. Harrington (Harvard University), and T. Therneau (Mayo Clinic)
- Regular - before April 15 $100 - after April 15 $135
Student or Retired $35