ROBUST MIXED EFFECTS MODEL FOR CLUSTERED FAILURE TIME DATA: APPLICATION TO HUNTINGTON'S DISEASE EVENT MEASURES

Area of Science:

• Biostatistics
• Statistical modeling
• Clinical research

Background:

• Clustered failure time data present challenges due to intraclass dependency and censoring.
• Existing methods often require restrictive modeling or distributional assumptions.
• Accurate modeling is crucial for reliable clinical and statistical research.

Purpose of the Study:

• To develop a novel, flexible approach for modeling clustered failure times with censoring.
• To address limitations of existing methods by avoiding restrictive assumptions.
• To provide new insights into event time differences in studies like Huntington's disease.

Main Methods:

• Utilized a logit transformation to link clustered failure time distributions with covariates and random effects.
• Employed pseudovalues to manage censored data and splines for functional covariate effects.
• Developed semiparametric techniques for fitting additive logistic mixed effects models without specifying random effect distributions.

Main Results:

• The proposed semiparametric methods provide consistent estimators regardless of random effect distribution or dependency structure.
• The approach demonstrated flexibility and robustness in theoretical and empirical evaluations.
• The method offers new insights into motor vs. cognitive impairment event times in Huntington's disease.

Conclusions:

• The novel method effectively handles clustered, censored failure time data without imposing restrictive assumptions.
• This approach enhances statistical modeling capabilities in clinical research.
• The application to Huntington's disease highlights the method's practical utility and potential for new discoveries.