Implementing RI-CLPM with Multiple ROIs in brms or blavaan for Neuroimaging Data

AFNI Message Board
, ,
1

Dear AFNI community,

I'm seeking advice on implementing a Random Intercept Cross-Lagged Panel Model (RI-CLPM) in R using either brms or blavaan. My specific use case involves neuroimaging data, and I haven't found examples that address my particular needs.

image
image706×648 50.7 KB

Here are the key points:

  1. X variable: A behavioral measure, potentially involving multiple indicators related to a latent factor at each wave.
  2. Y variable: Univariate BOLD% signal change.
  3. Multiple ROIs: The Y variable (BOLD signal) comes from multiple ROIs, which I'd like to model as a random factor, similar to how the RBA function in AFNI handles it (e.g., ... + (1 | ROI)).
  4. Time points: The model will involve multiple waves of data (e.g.: 2).

My questions are:

  1. Has anyone implemented an RI-CLPM model in brms or blavaan with BOLD signal from multiple ROIs as the outcome variable? Am I on the right track to want to this?
  2. How can I incorporate the random effect of ROIs into the model structure?
  3. Are there any specific considerations or challenges when using neuroimaging data in this type of model? I will run simulation to determine the sample size, but the goal is to use a large secondary dataset.
  4. Can you provide a basic code structure or example that I could adapt for this purpose?
  5. Are there any relevant papers or resources you'd recommend that combine RI-CLPM with neuroimaging data, particularly handling multiple ROIs?

Any guidance, code snippets, or references would be greatly appreciated. Thank you in advance for your help!

2

I don't have experience with cross-lagged multilevel panel models, nor do I have time to explore them at the moment. You might find some useful information on the Stan message board.

Gang Chen

1 Like