Know Your Trials
Patients battling rare orphan diseases often miss life-saving experimental treatments because matching complex clinical eligibility protocols manually is exceedingly tedious and fragmented. Built during HexaFalls, Know Your Trials pairs a modern neomorphic React interface with a FastAPI backend to extract patient phenotypes, validate conditions with Orphanet, and automatically identify eligible trials using vector retrieval.
Clinical Document Ingestion & BioBERT NER
Parses unstructured clinical case notes and medical PDF reports using PyPDF2 and BioBERT Named Entity Recognition (NER). Extracts granular clinical variables including patient age, genetic biomarkers, diagnostic indicators, and phenotypic manifestations to build a structured patient profile.
Orphanet Rare Disease Ontology Validation
Cross-references extracted candidate pathologies against the official Orphanet rare disease registry. Normalizes non-standard physician phrasing into standardized ORPHA codes, mitigating vocabulary drift and ensuring consistent eligibility indexing across international clinical trial registries.
LangGraph Multi-Agent Trial Matching Engine
Orchestrates a clinical retrieval pipeline using LangGraph and ChromaDB vector embeddings. Matches candidate patient profiles against active global trial criteria, performing automated reasoning over multi-clause inclusion and exclusion constraints before persisting patient audits in Supabase.