Podcast: BRAINIAC

The provided sources describe BRAINIAC, a novel Bayesian statistical model designed for neuroimaging research to analyze the complex relationships between whole-brain data (primarily resting-state fMRI) and cognitive or behavioral traits. This model addresses the challenges of replicating findings from smaller studies by simultaneously considering all brain features and assessing the contribution of predefined brain feature groupings called annotations. BRAINIAC estimates the total variance in a cognitive trait explained by brain data and identifies whether certain annotated feature groups are particularly enriched for these associations, as demonstrated in its application to the ABCD Study data for crystallized intelligence and psychopathology, with validation using the HCP-D dataset. The method aims to offer a more reliable and comprehensive understanding of brain-behavior links by moving beyond traditional single-feature or sparsity-assuming analyses.

**Generated by NotebookLM.

For more information:

brainiac_version_4_0.pdf
File Size:	845 kb
File Type:	pdf

Download File

---

WESLEY THOMPSON,
​​Principal Investigator