Bayesian semiparametric modeling for HIV longitudinal data with censoring and skewness

Area of Science:

• Biostatistics
• Longitudinal Data Analysis
• Biomedical Research

Background:

• Gaussian assumptions are common but often inadequate for skewed biomedical data.
• Clinical assays frequently involve censored responses (upper/lower limits).
• Longitudinal responses can exhibit nonlinear relationships with covariates like time.

Purpose of the Study:

• To develop a robust Bayesian semiparametric model for longitudinal censored data.
• To address skewness, censoring, and nonlinearity in biomedical response variables.
• To improve the analysis of complex patient trajectories in clinical studies.

Main Methods:

• Utilized a Bayesian semiparametric longitudinal censored model.
• Employed splines for approximating the general mean.
• Incorporated wavelets for modeling individual subject trajectories.
• Applied the skew-normal distribution for modeling random effects.

Main Results:

• The developed model effectively handles skewed and censored longitudinal data.
• Demonstrated accurate analysis of simulated data.
• Successfully applied to real-world AIDS/HIV viral load data.

Conclusions:

• The proposed Bayesian model offers a flexible and accurate approach for analyzing complex biomedical longitudinal data.
• This method overcomes limitations of traditional Gaussian-based models in the presence of skewness and censoring.
• The model provides enhanced insights into patient responses in clinical settings.