A semiparametric joint model for cluster size and subunit-specific interval-censored outcomes

Area of Science:

• Biostatistics
• Statistical Modeling
• Biomedical Data Analysis

Background:

• Clustered data are common in biomedical research, with observations within clusters being associated.
• Informative cluster size occurs when outcome variables correlate with the number of subunits per cluster.
• Existing marginal models for informative cluster size lack efficiency and cannot estimate within-cluster associations.

Purpose of the Study:

• To propose a semiparametric joint model for clustered, interval-censored event time data with informative cluster size.
• To address limitations of existing marginal approaches by enabling estimation of within-cluster associations.
• To develop a statistically robust and efficient method for analyzing complex biomedical clustered data.

Main Methods:

• Developed a semiparametric joint model incorporating a random effect to capture intra-cluster event time associations and event time-cluster size associations.
• Employed a sieve maximum likelihood approach for estimation.
• Designed a computationally efficient expectation-maximization algorithm for model implementation.

Main Results:

• The proposed estimators demonstrate strong consistency.
• Euclidean components of the estimators are asymptotically normal and achieve semiparametric efficiency.
• Simulation studies confirm the method's finite-sample performance, efficiency, and robustness.

Conclusions:

• The novel joint modeling approach effectively handles clustered interval-censored event time data with informative cluster size.
• The method provides consistent and efficient estimation of both marginal and within-cluster associations.
• The approach is validated through simulations and applied to a periodontal disease dataset, demonstrating its practical utility.