qEEG Council

6-stage LLM deliberation workflow for analyzing qEEG/ERP reports, with multi-model peer review and clinician portal publishing.

qEEG Council is a multi-model AI deliberation system for analyzing quantitative EEG and event-related potential reports. It orchestrates a 6-stage workflow across multiple LLM providers – OpenAI, Anthropic, and Google – via a CLIProxyAPI router. The deliberation pattern (initial analysis, anonymized peer review, revision, consolidation, final review vote, and publication-ready draft) produces higher-quality clinical reports than any single model pass.

Key Features

• 6-stage multi-model deliberation with peer review and revision cycles
• Multimodal analysis – vision-capable models receive PDF page images in Stage 1
• Patient management with report upload, OCR via Tesseract, and page extraction
• SSE-streamed real-time progress during analysis runs
• Explainer video pipeline integration with clinician portal publishing
• Markdown and PDF export for the final consolidated report

Technical Architecture

The backend is FastAPI with SQLite for patient and report management. PDF uploads are processed with Tesseract OCR for text extraction and page image rendering for vision-capable models. The 6-stage pipeline routes through multiple LLM providers, with each model producing independent analysis before the anonymized peer review stage.

The consolidation stage synthesizes all model revisions into a unified report, and a final review vote determines publication readiness. The React/Vite frontend provides real-time progress via server-sent events. The platform also integrates with an explainer video pipeline for patient education, with QC verification against council artifacts and automatic publishing to a clinician portal sync folder.

Released under a noncommercial license.