Related Experiment Videos
Nonparametric regression estimates with censored data: local linear smoothers and their applications
1Department of Biostatistics, Graduate School of Public Health, University of Pittsburgh, Pennsylvania 15261, USA. kim@toutatis.pci.upmc.edu
Biometrics
|January 12, 1999
Summary
This study enhances nonparametric regression for censored survival data by using local linear fits to improve bias and boundary effects. The new method accurately estimates conditional survival, hazard, mean, and median functions.
Area of Science:
- Statistics
- Biostatistics
- Survival Analysis
Background:
- Nonparametric regression is used for censored time-to-event data.
- Kernel-based methods estimate conditional survival functions but have bias and boundary issues.
Purpose of the Study:
- To improve existing kernel-based nonparametric regression techniques.
- To address bias and boundary effects in estimating conditional survival functions for censored data.
Main Methods:
- Local linear regression fits are proposed to enhance kernel-based methods.
- The improved procedure is demonstrated using simulated and real-world censored survival data.
Main Results:
- Local linear fits effectively reduce bias and boundary effects.
- Consistency is established for estimates of conditional survival, cumulative hazard, mean, and median functions.
Conclusions:
- The local linear approach offers an improvement over traditional kernel methods for censored survival data.
- This method provides reliable estimates for key survival function components.