Frontend
Next.js 16 App Router, React 19, TypeScript, Tailwind CSS v4, and shadcn/ui for fast clinical UI iteration.
Technology behind Synth
This page outlines the stack, flow, and privacy controls used to convert visit audio into structured clinical outputs and patient-safe interactions.
Core Data Layer
Prisma ORM with PostgreSQL schema for users, patients, visits, notes, share links, appointments, and care plan items.
AI Layer
Amazon Nova (via Amazon Bedrock) powers conversation summaries, SOAP draft generation, and grounded chat responses.
AWS Runtime
AWS-hosted deployment uses ECS Fargate, RDS PostgreSQL, Bedrock, S3, and CloudWatch for a clean full-stack runtime.
How it works
1. Capture
Browser audio capture starts in Transcribe. Session state is tracked in real time with start, pause, and stop controls.
2. Transcribe
Transcript text (or browser live transcript) is normalized into speaker-labeled segments for clinician and patient dialogue.
3. Structure
Saving runs /api/transcribe/save to generate summary + SOAP notes and persist visit-linked documentation.
4. Deliver
Clinician and patient surfaces consume the same visit context so guidance stays grounded in stored visit evidence.
Privacy and security
- Credential authentication with NextAuth and bcrypt password hashing.
- Role-based API guards so clinician-only workflows stay access controlled.
- Tokenized share links for patient views with revocation support in data model.
- HTTPS/TLS for data in transit between browser and server.
- AES-256 encryption at rest via managed infrastructure configuration (recommended deployment baseline).
- Clear auditability path through visit-linked records and AWS CloudWatch logging.
Deployment note: production healthcare environments should pair this application with compliance controls (BAA-ready vendors, key management, logging retention, and access governance) before handling regulated PHI.