NeuroMambaLLM: Dynamic Graph Learning of fMRI Functional Connectivity in Autistic Brains Using Mamba and Language Model Reasoning

Researchers unveiled NeuroMambaLLM, a new AI that fuses Mamba's state-space models with frozen LLMs to analyze dynamic brain connectivity from fMRI scans. It learns adaptive, latent graphs from raw BOLD time-series data, capturing transient neural patterns missed by static methods. The system reportedly achieves over 90% accuracy in autism diagnosis and can generate clinical reports, moving beyond simple classification to reasoning about brain activity over time.