Bayesian inference for joint modelling of longitudinal continuous, binary and ordinal events.

Area of Science:

• Biostatistics
• Longitudinal Data Analysis
• Multivariate Statistics

Background:

• Medical studies often collect repeated measurements of continuous, binary, and ordinal outcomes from the same patient.
• Existing methods for joint analysis are limited, particularly for trivariate (three-variable) longitudinal responses.
• There is a need for robust statistical methods to analyze the association between different types of health outcomes.

Purpose of the Study:

• To propose and evaluate a novel method for jointly modeling trivariate longitudinal responses (continuous, binary, and ordinal).
• To account for the inherent associations between different health outcomes in statistical modeling.
• To improve the efficiency and reliability of statistical inferences in medical research.

Main Methods:

• Development of conditional joint random-effects models for trivariate longitudinal data.
• Application of Bayesian analysis methods for statistical inference.
• Validation through simulation studies and analysis of a real-world cohort study from North West England.

Main Results:

• Joint modeling of trivariate outcomes resulted in smaller standard deviations and more stable parameter estimates compared to separate analyses.
• The proposed joint analysis demonstrated improved efficiency and reliability in statistical inferences.
• In real data analysis, the joint model yielded a significantly smaller deviance information criterion value than separate models.

Conclusions:

• Jointly modeling continuous, binary, and ordinal longitudinal outcomes is statistically advantageous.
• The proposed conditional joint random-effects models provide a powerful tool for analyzing complex health data.
• This approach enhances the precision and trustworthiness of statistical inferences in medical research.