Piecewise mixed-effects models with skew distributions for evaluating viral load changes: A Bayesian approach.

Area of Science:

• Biostatistics
• Epidemiology
• Infectious Diseases

Background:

• Human immunodeficiency virus (HIV) dynamics are crucial for evaluating antiretroviral (ARV) therapies in acquired immuno deficiency syndrome (AIDS) research.
• Viral load changes in plasma are key indicators of ARV potency in clinical trials.
• Subject viral load trajectories often exhibit 'broken stick'-like dynamics, signifying multiple phases of decline and increase.

Purpose of the Study:

• To propose novel statistical models for analyzing longitudinal viral load data.
• To address limitations of traditional normal distribution assumptions in modeling within- and among-subject variations.
• To apply these models within a Bayesian framework for analyzing HIV/AIDS patient data.

Main Methods:

• Development of piecewise linear mixed-effects models incorporating skew-elliptical distributions.
• Application of a Bayesian framework for statistical inference.
• Analysis of real-world viral load data from an acquired immuno deficiency syndrome (AIDS) study.
• Comparison of various candidate models to assess model performance.

Main Results:

• The proposed models effectively describe the time trend of viral load, capturing complex dynamics.
• Analysis revealed that assuming a skew distribution is vital for reliable results, especially with skewed data.
• Biologically significant findings were derived from the viral load data analysis.

Conclusions:

• Piecewise linear mixed-effects models with skew-elliptical distributions offer a robust approach for longitudinal data analysis in HIV/AIDS research.
• The choice of distribution (skew vs. normal) significantly impacts the reliability of findings in viral load studies.
• This methodology provides valuable insights into treatment effectiveness and disease progression.