Medical AI document analysis
Amorphous.health
An AI-powered interface that reads medical documents, surfaces clinical terms automatically and lets practitioners explore findings without ever leaving the page.
The brief
THE CHALLENGE
A health-tech startup had a working Python NLP model that could extract and classify medical terminology from clinical documents - but no way to put it in front of practitioners. They needed a frontend that could ingest raw PDFs, communicate with the model and present results in a way clinicians would trust.
How we did it
OUR APPROACH
PDF ingestion layer
Built a JavaScript pipeline that reads uploaded PDFs client-side, extracts the full text content and prepares it for API submission - without any file ever leaving the browser.
NLP API integration
Integrated the existing Python model as an API endpoint. The frontend sends extracted content, the model returns a structured map of flagged medical terms with classifications.
Custom PDF renderer
Developed a bespoke rendering library that overlays the original PDF with colour-coded highlights - one distinct colour per clinical category - directly on the document as the user reads.
Inline term explorer
Any highlighted term can be clicked to open an inline panel showing its medical classification, category and contextual notes - no page navigation required.
The outcome
RESULTS THAT MOVED
What we shipped
DELIVERABLES
A complete package, from first sketch through live, measured and growing. Want the same for your project?
Book a discovery callNext project