"Smooth" semiparametric regression analysis for arbitrarily censored time-to-event data

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Time-to-event data analysis often involves censoring, complicating standard regression models.
• Popular semiparametric models leave key components, like baseline hazard functions, unspecified.
• Existing methods may struggle with arbitrary censoring patterns.

Purpose of the Study:

• To propose a general regression framework for time-to-event data with arbitrary censoring.
• To provide a unified approach for various survival models, including proportional hazards, proportional odds, and accelerated failure time models.
• To enable principled model selection and facilitate extensions.

Main Methods:

• Utilizes a truncated series expansion to approximate unspecified density functions.
• Employs likelihood-based inference with adaptive truncation degree selection for flexibility.
• Applies the framework to data with any censoring pattern.

Main Results:

• Demonstrates computational and conceptual simplicity for likelihood inference.
• Shows the framework accommodates and unifies popular survival regression models.
• Validates the approach through simulations and real-world data applications.

Conclusions:

• The proposed framework offers a flexible and unified approach to survival data analysis.
• It simplifies the handling of arbitrary censoring patterns and model selection.
• The method is robust and applicable to diverse time-to-event studies.