A flexible, computationally efficient method for fitting the proportional hazards model to interval-censored data.

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• The proportional hazards (PH) model is widely used for time-to-event data analysis.
• Analyzing interval-censored data under the PH model presents significant challenges with existing methods.

Purpose of the Study:

• To present a novel method for analyzing interval-censored data within the PH model framework.
• To enhance flexibility and parameter estimation in survival analysis for interval-censored data.

Main Methods:

• Utilized a monotone spline representation to approximate the cumulative baseline hazard function.
• Developed a novel expectation-maximization (EM) algorithm with two-stage data augmentation using latent Poisson variables.
• Implemented a method for estimating parameters and providing closed-form variance estimates.

Main Results:

• The proposed approach offers substantial modeling flexibility with a finite number of parameters.
• The developed EM algorithm demonstrates ease of implementation, robustness, and rapid convergence.
• The methodology provides closed-form variance estimates for improved statistical inference.

Conclusions:

• The new method effectively addresses the challenges of analyzing interval-censored data under the PH model.
• The approach is validated through simulation studies and application to a National Cancer Institute randomized trial.
• This technique offers a robust and efficient alternative for survival data analysis.