Spaces:
Running
Running
DeepCritical Documentation
Medical Drug Repurposing Research Agent
AI-powered deep research system for accelerating drug repurposing discovery.
Quick Links
Architecture
- Overview - Project overview, use case, architecture, timeline
- Design Patterns - 17 technical patterns, reference repos, judge prompts, data models
Implementation (Start Here!)
- Roadmap - Phased execution plan with TDD
- Phase 1: Foundation β - Tooling, config, first tests
- Phase 2: Search β - PubMed search
- Phase 3: Judge β - LLM evidence assessment
- Phase 4: UI β - Orchestrator + Gradio
- Phase 5: Magentic β - Multi-agent orchestration
- Phase 6: Embeddings β - Semantic search + dedup
- Phase 7: Hypothesis β - Mechanistic reasoning
- Phase 8: Report β - Structured scientific reports
- Phase 9: Source Cleanup π - Remove DuckDuckGo
- Phase 10: ClinicalTrials π - Clinical trials API
- Phase 11: bioRxiv π - Preprint search
Guides
- Setup Guide (coming soon)
- Deployment Guide - Gradio, MCP, and Modal launch steps
Development
- Testing Strategy - Unit, Integration, and E2E testing patterns
- Contributing (coming soon)
What We're Building
One-liner: AI agent that searches medical literature to find existing drugs that might treat new diseases.
Example Query:
"What existing drugs might help treat long COVID fatigue?"
Output: Research report with drug candidates, mechanisms, evidence quality, and citations.
Architecture Summary
User Question β Research Agent (Orchestrator)
β
Search Loop:
β Tools (PubMed, Web Search)
β Judge (Quality + Budget)
β Repeat or Synthesize
β
Research Report with Citations
Hackathon Tracks
| Track | Status | Key Feature |
|---|---|---|
| Gradio | β Planned | Streaming UI with progress |
| MCP | β Planned | PubMed as MCP server |
| Modal | π Stretch | GPU inference option |
Team
- Physician (medical domain expert) β
- Software engineers β
- AI architecture validated by multiple agents β
Status
| Phase | Status |
|---|---|
| Phases 1-8 | β COMPLETE |
| Phase 9: Remove DuckDuckGo | π SPEC READY |
| Phase 10: ClinicalTrials.gov | π SPEC READY |
| Phase 11: bioRxiv | π SPEC READY |
Architecture Review: PASSED (98-99/100) Phases 1-8: COMPLETE Next: Phases 9-11 (Multi-Source Enhancement)