CyberNeuro: A Privacy-Preserving Agentic Workbench for Cohort-Scale Neuroimage and Clinical Data Analysis
3.60T1 sourcearXiv cs.MA
Source record
Published by arXiv cs.MA (T1 source). The original is at https://arxiv.org/abs/2607.28841.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryCyberNeuro is an agentic workbench for neuroimaging and clinical data analysis, built on four LLM agents (Planner, Validator, Dispatcher, Reporter) communicating via a secure MCP bridge. It uses a local model called WandaMind to preserve data privacy, supports natural-language workflows, and raises NeuroBench held-out accuracy from 40% to 69%.
Why it mattersDeployable four-agent architecture with a local LLM for privacy-sensitive domains; concrete accuracy and token-cost numbers against a baseline, not a marketing claim.
Cited by
No citations on record.
