Modeling Alzheimer's Disease Biomarkers' Trajectory in the Absence of a Gold Standard Using a Bayesian Approach

Area of Science:

• Neurology
• Biostatistics
• Biomarker Research

Background:

• Alzheimer's Disease (AD) research increasingly focuses on preclinical biomarkers.
• Current methods often rely on clinical diagnoses, potentially missing early disease stages.
• Existing frameworks map biomarker progression but may be limited by diagnostic proxies.

Purpose of the Study:

• To develop a novel Bayesian approach for modeling latent Alzheimer's Disease (AD) status.
• To analyze biomarker trajectories as nonlinear functions of disease progression.
• To reduce bias from clinical diagnoses and better understand early AD biomarker evolution.

Main Methods:

• Developed a Bayesian model to directly estimate underlying AD status as a latent process.
• Modeled biomarker trajectories as nonlinear functions of latent disease progression.
• Incorporated subject-specific latent trajectories and random intercepts to handle heterogeneity.

Main Results:

• Simulation studies confirmed the model's performance.
• Application to the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset provided interpretable clinical insights.
• Demonstrated the model's capability in understanding AD biomarker evolution across the continuum.

Conclusions:

• The novel Bayesian approach offers a data-driven method for exploring AD progression.
• This method can reduce bias associated with clinical diagnoses in early AD research.
• The findings facilitate a deeper understanding of Alzheimer's Disease biomarker changes over time.