Reproduce “SABER: A Semantic-Aligned Brain Network Analysis Framework via Multi-scale Hypergraphs” 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
Yidan Xu, Xiangmin Han, Rundong Xue, Huihui Ye
The SABER framework integrates clinical semantic knowledge from Large Language Models directly into brain network classification for neurodevelopmental disorder diagnosis. It achieves superior diagnostic accuracy and stability on ABIDE I and ADHD-200 datasets by modeling multi-scale, high-order brain interactions and actively aligning these with patient- and disease-specific semantic priors.
Key insights
- Directly integrating LLM-derived semantic information at the decision-making level significantly enhances brain network classification performance and stability compared to using semantics as auxiliary features.
- Multi-scale hypergraph modeling effectively captures complex high-order interactions and hierarchical functional organization within brain networks, crucial for robust disorder diagnosis.
- A unified attention mechanism across multi-scale hypergraphs, coupled with adaptive fusion and a decision-level semantic query, is critical for coherently integrating diverse brain network patterns and patient-specific semantic cues.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.