A flexible AFT model for misclassified clustered interval-censored data

Area of Science:

• Biostatistics
• Survival Analysis
• Longitudinal Data Analysis

Background:

• Longitudinal studies often involve clustered time-to-event data.
• Data can be interval-censored, with events determined within examination intervals.
• Event occurrence assessment may be subject to misclassification.

Purpose of the Study:

• To propose a flexible statistical modeling approach for clustered, interval-censored time-to-event data with misclassification.
• To develop a model that accounts for potential variations in event assessment by different examiners.
• To estimate underlying event time distributions and misclassification parameters without external information.

Main Methods:

• Utilized an accelerated failure time model with random effects for clustered time-to-event data.
• Employed a penalized Gaussian mixture model for random effects to avoid restrictive distributional assumptions.
• Developed a general misclassification model and a Bayesian implementation for parameter estimation.

Main Results:

• Empirical evidence demonstrates the model's capability to estimate time-to-event distributions and misclassification parameters.
• The proposed method does not require external information for estimating misclassification parameters.
• Simulation studies evaluated the impact of ignoring misclassification in clustered time-to-event data analysis.

Conclusions:

• The proposed flexible modeling approach effectively handles clustered, interval-censored time-to-event data with misclassification.
• The Bayesian implementation allows for robust estimation of key parameters, including misclassification.
• Neglecting misclassification can significantly affect the analysis of such data, highlighting the model's importance.