Causal functional mediation analysis with an application to functional magnetic resonance imaging data.

Area of Science:

• Neuroscience
• Statistics
• Data Science

Background:

• Task-based functional magnetic resonance imaging (fMRI) aims to quantify effective brain connectivity during stimuli presentation.
• Assessing the dynamic changes in effective connectivity is crucial for understanding brain function.
• Causal mediation analysis is a common tool for exploring mechanisms between stimuli and brain activation, but its application to continuous functional data is limited.

Purpose of the Study:

• To extend causal mediation analysis to functional data, specifically for scenarios with continuous treatment, mediator, and outcome.
• To introduce semiparametric functional linear structural equation models for analyzing dynamic brain connectivity.
• To enable the estimation of individual effect curves in functional data analysis.

Main Methods:

• Development of semiparametric functional linear structural equation models.
• Discussion of causal assumptions pertinent to functional data mediation analysis.
• Application of the proposed models to task-based fMRI data.

Main Results:

• The study successfully applies causal mediation analysis to functional data.
• The developed models allow for the estimation of individual effect curves, providing detailed insights.
• The application to fMRI data offers a novel perspective on dynamic brain connectivity.

Conclusions:

• The proposed semiparametric functional linear structural equation models extend causal mediation analysis to continuous functional data.
• This approach provides a new method for studying dynamic brain connectivity in task-based fMRI.
• An R package (cfma) is available for implementing these novel statistical models.