Nonparametric estimation of the conditional mean residual life function with censored data

Area of Science:

• Statistics
• Survival Analysis
• Reliability Engineering

Background:

• The conditional mean residual life (MRL) function estimates remaining lifetime based on survival time and covariates.
• It is crucial in reliability and actuarial science, especially for right-tail distribution analysis.
• Existing semi-parametric MRL models have theoretical limitations.

Purpose of the Study:

• To propose and evaluate novel nonparametric methods for estimating the conditional MRL function.
• To address theoretical shortcomings of current semi-parametric MRL models.
• To compare the performance of proposed methods against existing techniques.

Main Methods:

• Development of two new nonparametric estimators for the conditional MRL function.
• Establishment of asymptotic properties, including consistency and normality.
• Empirical investigation using Monte Carlo simulations and bootstrap confidence intervals.

Main Results:

• The proposed nonparametric estimators demonstrate reliable performance.
• Comparison with popular semi-parametric methods shows advantages in certain data scenarios.
• Asymptotic properties of the new estimators are theoretically validated.

Conclusions:

• The novel nonparametric methods provide a robust alternative for conditional MRL estimation.
• These methods offer valuable insights in reliability and actuarial studies.
• The approach is successfully applied to lung cancer survival data.