Related Experiment Videos
Marginal regression of gaps between recurrent events.
1Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA, USA. eugene@fhcrc.org
Lifetime Data Analysis
|December 3, 2003
Summary
This study introduces a new marginal proportional hazards model for analyzing recurrent event data, accounting for intra-individual correlation. The proposed method effectively estimates covariate effects in clustered survival data, demonstrating practical utility with bladder tumor data.
Area of Science:
- Biostatistics
- Survival Analysis
- Statistical Modeling
Background:
- Recurrent event data often show intra-individual correlation due to observed covariates and random effects.
- Populations can be modeled as mixtures of individual renewal processes, with interest in covariate effects.
Purpose of the Study:
- To propose and investigate a marginal proportional hazards model for recurrent event gap times.
- To develop a general inference procedure for clustered survival data with informative cluster size.
Main Methods:
- A marginal proportional hazards model for recurrent event gaps.
- Establishing a link between gap times and clustered survival data.
- A novel inference procedure based on functional Cox regression.
Main Results:
- Large-sample theory is established for the proposed estimators.
- Numerical studies confirm the procedure's good performance with practical sample sizes.
- The method is illustrated using bladder tumor data.
Conclusions:
- The proposed marginal proportional hazards model and inference procedure are effective for recurrent event data.
- The methodology addresses intra-individual correlation and informative cluster size in survival analysis.
- The approach offers a valuable tool for analyzing complex recurrent event data in various applications.