Semiparametric regression analysis of interval-censored competing risks data

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Competing risks data are common in medical studies, but often failure times are only known to lie within intervals.
• Existing methods may not fully capture complex covariate effects or handle missing failure cause information.

Purpose of the Study:

• To develop flexible semiparametric regression models for interval-censored competing risks data.
• To estimate the effects of time-varying covariates on cumulative incidence functions.
• To accommodate arbitrary examination times and missing failure cause data.

Main Methods:

• Formulated semiparametric regression models for sub-distribution functions.
• Employed nonparametric maximum likelihood estimation.
• Developed a fast and stable EM-type algorithm for computation.
• Utilized modern empirical process theory to establish theoretical properties of estimators.

Main Results:

• The proposed estimators are consistent, asymptotically normal, and semiparametric efficient.
• Extensive simulations demonstrate good performance in realistic scenarios.
• The methods successfully analyzed HIV-1 data with varying viral subtypes.

Conclusions:

• The developed statistical framework provides a robust tool for analyzing complex interval-censored competing risks data.
• This approach enhances the understanding of disease progression and covariate effects in medical research.
• The methods are applicable to various biomedical studies with similar data structures.