A Joint Bayesian Model for Change-Points and Heteroskedasticity Applied to the Canadian Longitudinal Study on Aging

Area of Science:

• Biostatistics
• Physiology
• Biomarker Discovery

Background:

• Homeostasis, maintaining stable internal physiological parameters, is crucial for health.
• Deviations from optimal physiological levels ('sweet spots') can indicate disease.
• Identifying biomarkers with sweet spots traditionally requires separate change-point and heteroskedasticity analyses, risking inaccuracy.

Purpose of the Study:

• To develop a unified Bayesian framework for detecting biomarkers with 'sweet spots'.
• To improve the accuracy and robustness of identifying physiological sweet spots.
• To quantify heteroskedasticity and its effects in regression models.

Main Methods:

• Proposed Bayesian Testing for Heteroskedasticity and Sweet Spots (BTHS) framework.
• Integrated sampling-based parameter estimation and Bayes factor computation.
• Applied BTHS to analyze blood elements from the Canadian Longitudinal Study on Aging.

Main Results:

• BTHS provides a unified approach, eliminating the need for separate analyses.
• The method offers detailed insights into heteroskedasticity magnitude and shape.
• Identified nine blood elements exhibiting significant sweet spot variance effects.

Conclusions:

• BTHS enables robust identification of physiological sweet spots without strong assumptions.
• The framework enhances change-point detection and heteroskedasticity testing.
• This approach advances biomarker discovery for health and disease monitoring.