The Challenge
After surgery, patients and caregivers have questions. Lots of questions. But reaching their actual care team is hard, and generic FAQ pages don’t address their specific situations.
What if patients could get answers grounded in what their doctor actually said—indexed and searchable from their consultation recordings?
My Role
Technical co-founder. Led a team of 3 iOS engineers. Built the clinical NLU pipeline and conversational retrieval system.
The Approach
We built a proprietary clinical NLU service:
- Speaker diarization to separate doctor from patient in consultation recordings
- Semantic indexing of the doctor’s explanations
- Retrieval-based Q&A letting patients ask questions and hear answers in context—grounded in what their doctor actually said
Results
- Deployed and evaluated at Seattle Children’s Hospital and Virginia Mason Hospital