Reproduce “Bayesian recurrent state space model for rs-fMRI” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Arunesh Mittal, Scott Linderman, John Paisley, Paul Sajda
Researchers at Columbia and Stanford Universities developed a hierarchical Bayesian recurrent state space model for resting-state fMRI, which improved classification accuracy for Mild Cognitive Impairment (MCI) to 79.17% compared to 72.58% from a deep learning baseline. The model also unveiled distinct class-specific and shared functional connectivity patterns, offering interpretable insights into neuropathologies.
Key insights
- Modeling brain dynamics with shared spatial covariance structures but class-conditional temporal transitions allows for differentiating common and disease-specific network alterations.
- Explicitly accounting for recurrent state transitions provides a more accurate representation of dynamic brain activity in rs-fMRI data.
- The hierarchical Bayesian framework enables direct comparison of latent structures across different disease classes, offering interpretable insights into neuropathologies.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.