Semiparametric Regression Analysis of Multiple Right- and Interval-Censored Events

Area of Science:

• Biostatistics
• Epidemiology
• Health Sciences Research

Background:

• Health research frequently encounters right-censored (end of follow-up) and interval-censored (periodic examinations) events.
• Distinguishing between symptomatic and asymptomatic disease detection presents unique analytical challenges.

Purpose of the Study:

• To develop a statistical framework for jointly modeling multiple right- and interval-censored events.
• To incorporate time-dependent covariates and random effects to capture event dependencies.
• To enable dynamic prediction of disease incidence based on evolving patient data.

Main Methods:

• Semiparametric proportional hazards models with random effects were formulated.
• Nonparametric maximum likelihood estimation was employed.
• A stable Expectation-Maximization (EM) algorithm was developed for computational efficiency.

Main Results:

• The proposed estimators were found to be consistent.
• Parametric components demonstrated asymptotic normality and efficiency.
• A reliable method for estimating the covariance matrix was established using profile likelihood or nonparametric bootstrap.

Conclusions:

• The joint modeling approach effectively handles complex censoring patterns in health data.
• The method provides accurate dynamic predictions for disease incidence.
• The developed statistical techniques offer robust tools for epidemiological studies.