A Bayesian transformation model for informative partly interval-censored data with covariates subject to measurement...

Area of Science:

• Statistics
• Biostatistics
• Survival Analysis

Background:

• Linear transformation models are widely used for failure time data regression.
• Existing methods struggle with covariates having measurement error and informative censoring.
• Ignoring these issues can cause biased estimates and incorrect conclusions in survival analysis.

Purpose of the Study:

• To develop a robust statistical model for analyzing failure time data with both covariate measurement error and informative (partly interval) censoring.
• To propose a flexible Bayesian estimation procedure to address these complex data scenarios.
• To provide a reliable method for regression analysis in challenging survival data settings.

Main Methods:

• A novel joint Bayesian model is proposed for regression analysis.
• I-splines are utilized to approximate unknown functions within the model.
• A stable Markov chain Monte Carlo (MCMC) algorithm with four-stage data augmentation is developed for implementation.

Main Results:

• Extensive simulation studies demonstrate the effectiveness of the proposed Bayesian method.
• The Bayesian approach significantly outperforms naive methods by mitigating bias.
• The method is shown to be effective in handling complex censoring and measurement error scenarios.

Conclusions:

• The proposed Bayesian joint model provides a powerful and flexible tool for survival data analysis.
• The method accurately accounts for covariate measurement errors and informative censoring.
• This approach offers a reliable alternative for regression analysis in complex failure time data.