Related Experiment Videos
A local likelihood proportional hazards model for interval censored data
Rebecca A Betensky1, Jane C Lindsey, Louise M Ryan
1Department of Biostatistics, Harvard School of Public Health, 655 Huntington Avenue, Boston, Massachusetts 02115, USA. betensky@hsph.harvard.edu
Statistics in Medicine
|January 10, 2002
Summary
Local likelihood methods offer a flexible approach to proportional hazards regression for censored survival data. This technique provides interpretable baseline hazard functions and covariate effects, enhancing survival analysis.
Area of Science:
- Biostatistics
- Survival Analysis
- Statistical Modeling
Background:
- Proportional hazards models are widely used for analyzing time-to-event data.
- Handling right and interval censored data presents unique statistical challenges.
- Existing methods may lack flexibility in modeling the baseline hazard function.
Purpose of the Study:
- To introduce and evaluate local likelihood methods for fitting proportional hazards regression models.
- To accommodate arbitrary, smoothed baseline hazard functions in survival analysis.
- To provide interpretable baseline hazard estimates alongside covariate effects.
Main Methods:
- Application of local likelihood estimation techniques.
- Extension of the modified Expectation-Maximization (EM) algorithm.
- Fitting proportional hazards regression models to right and interval censored data.
Main Results:
- The proposed method yields an interpretable smoothed baseline hazard function.
- Estimates of global covariate effects are obtained.
- Demonstrated utility in analyzing breast cosmesis deterioration and HIV-1 infection rates.
Conclusions:
- Local likelihood methods provide a powerful and flexible tool for proportional hazards regression.
- The extended EM algorithm facilitates robust estimation for censored data.
- The approach is applicable to diverse biomedical and epidemiological datasets.