MELODY: Mediation Analysis in Logistic Regression for High-Dimensional Mediators and a Binary Outcome

Area of Science:

• Biostatistics
• Genomics
• Epidemiology

Background:

• Mediation analysis is crucial for understanding indirect effects of exposures on disease via omics data.
• Traditional methods struggle with high-dimensional mediators, binary outcomes, and non-rare diseases.

Purpose of the Study:

• To develop a novel mediation analysis framework (MELODY) for high-dimensional omics data and binary outcomes.
• To accurately quantify total mediation effects, addressing limitations of existing methods.

Main Methods:

• Developed MELODY (MEdiation analysis in LOgistic regression for high-Dimensional mediators and binarY outcome).
• Utilized a second-moment-based measure for total mediation effect quantification.
• Implemented a variable selection procedure for high-dimensional data bias reduction.

Main Results:

• MELODY demonstrates superior performance in simulations with non-rare diseases and high-dimensional mediators.
• The framework effectively quantifies total mediation effects, even with opposing mediator effects.

Conclusions:

• MELODY offers a robust approach for mediation analysis in complex, high-dimensional omics studies.
• Applied to Framingham Heart Study data, MELODY analyzed sex-based mediation pathways to coronary heart disease.