How anhsirksai/off-label-radar works
Off-Label Radar finds active clinical trials that test approved drugs beyond their labeled indications, using Jev’s typed decisions to score and rank evidence. Users enter drug names or a plain-language condition in the web UI; the API starts the corresponding workflow, which gathers registry, label, and literature data, then presents ranked candidates and supporting evidence. The README documents capabilities beyond the sampled excerpts; those links are included where documented, while unsampled implementation details are not assumed.
Main entry points: pipeline.py Drug workflow · condition.py Condition workflow · main.py FastAPI · db.py SQLite cache
Components 16
Research workflows
- Drug workflow
pipeline.py - Condition workflow
condition.py - Off-label gate
pipeline.py - Evidence lookup
pipeline.py - Judge and rank
pipeline.py
Data and judgment services
- ClinicalTrials.gov
- RxNorm
- DailyMed
- PubMed
- Jev API
- Request controls
net.py
Run state and storage
- SQLite cache
db.py - Run status
net.py
Other components
- Researcher
Connections 29
- Researcher → Radar UI: enters query
README.md - Radar UI → FastAPI: calls API
static/app.js - FastAPI → Drug workflow: starts drug run
main.py - FastAPI → Condition workflow: starts condition run
main.py - FastAPI → SQLite cache: reads candidates
main.py - FastAPI → Run status: reports status
main.py - Drug workflow → RxNorm: resolves drugs
README.md - Drug workflow → DailyMed: fetches labels
pipeline.py - Drug workflow → ClinicalTrials.gov: finds active trials
README.md - Condition workflow → ClinicalTrials.gov: searches condition
README.md - Condition workflow → Jev API: matches MeSH
condition.py - Condition workflow → RxNorm: resolves interventions
condition.py - Condition workflow → DailyMed: fetches labels
README.md - Condition workflow → Off-label gate: uses shared gate
README.md - Drug workflow → Off-label gate: judges trial pairs
pipeline.py - Off-label gate → Jev API: asks typed questions
pipeline.py - Off-label gate → SQLite cache: saves gate scores
pipeline.py - Drug workflow → Evidence lookup: requests evidence
README.md - Evidence lookup → PubMed: fetches abstracts
pipeline.py - Evidence lookup → SQLite cache: stores evidence
README.md - Drug workflow → Judge and rank: judges and ranks
README.md - Condition workflow → Judge and rank: uses shared judge
README.md - Judge and rank → Jev API: scores plausibility
README.md - Judge and rank → SQLite cache: persists results
README.md - Drug workflow → SQLite cache: caches workflow data
README.md - Condition workflow → SQLite cache: caches condition data
README.md - Drug workflow → Request controls: uses rate limits
pipeline.py - Condition workflow → Request controls: uses rate limits
condition.py - Radar UI → FastAPI: displays results
README.md