Partially linear mixed-effects joint models for skewed and missing longitudinal competing risks outcomes.

Area of Science:

• Biostatistics
• Clinical Data Analysis
• Longitudinal Studies

Background:

• Longitudinal competing risks data are common in clinical research.
• Skewness and missing data are frequent challenges in these datasets.
• Existing joint models often fail to address these data complexities.

Purpose of the Study:

• To propose novel partially linear mixed-effects joint models.
• To analyze longitudinal competing risks data with skewness and missingness.
• To provide a robust statistical framework for complex clinical data.

Main Methods:

• Utilized asymmetric distributions to handle skewness in model errors.
• Employed an informative missing data model to address missingness.
• Developed joint models integrating longitudinal, competing risks, and missing data processes.
• Applied a fully Bayesian approach for parameter estimation.

Main Results:

• The proposed models successfully analyzed skew longitudinal competing risks data with missingness.
• Implementation in an AIDS clinical study yielded significant findings.
• Simulation studies validated the effectiveness and accuracy of the proposed method.

Conclusions:

• The developed partially linear mixed-effects joint models offer a powerful tool for analyzing complex longitudinal competing risks data.
• The approach effectively accounts for skewness and missingness, enhancing analytical capabilities in clinical research.
• This methodology provides a more reliable framework for understanding disease progression and treatment outcomes in the presence of data complexities.