AI referral processing for a hospital network
Referrals from hospitals, clinics and physicians arrived by fax as TIFF or PDF files, and staff had to read, sort and key each one in by hand. We architected an event-driven, AI-powered pipeline that turns each incoming fax into a structured referral ready for the care team.
Documents are split into pages, normalized and merged into searchable PDFs. Queue-based workers then run OCR, classify and extract the content, predict outcomes and map the data to existing patient and provider records. Daily and per-user reports give managers real-time visibility.
Pipeline
- Ingest fax documents (TIFF/PDF)
- Split, convert and merge into searchable PDF
- OCR with Microsoft document AI
- Classify, extract and predict
- Map to existing records and create the referral
- Real-time UI updates and daily reporting
Our role
Solution architecture, research and POCs, hands-on development, code reviews and team enablement.