2025 · Healthcare AI
MediVoice
Clinical consultation transcription with medical NER for symptoms, diagnoses, and medications.
3clinical entity types extracted: symptoms, diagnoses, medications
- My role
- Designed and built the app, including the ASR pipeline and the medical NER layer.
- Result
- Consultations are transcribed in real time, with symptoms, diagnoses and medications pulled out automatically.
- Built with
- TypeScript, ASR, NER, Next.js
Problem
Doctors lose consultation time to note-taking, and general-purpose speech models stumble over clinical vocabulary.
My role
Designed and built the app, including the ASR pipeline and the medical NER layer.
What I built
- Real-time consultation transcription
- Medical NER for symptoms, diagnoses, and medications
- Domain-adapted vocabulary handling
Outcome
Consultations are transcribed in real time, with symptoms, diagnoses and medications pulled out automatically. The app is live.