CareBridge AI

CareBridge AI is a longitudinal and multimodal health intelligence layer that connects the patient and the clinician.

Comment

CareBridge AI

Inspiration

Clinicians spend significant time converting scattered notes, voice input, and patient reports into structured documentation. Patients should be able to speak or write naturally while the healthcare provider needs structured, meaningful SOAP notes without losing traceability or trust. CareBridge AI connects these workflows. It transforms patient-reported voice and text into structured longitudinal data, and turns that into cited, auditable clinical documentation.

Our key differentiator is a RAG pipeline built on a custom knowledge base grounded in medical books.

What It Does

AI Health Diary (Patient Side)

Patients log symptoms via voice or text in a mobile app.

  • Azure Speech-to-Text transcribes voice
  • Azure OpenAI extracts structured symptoms such as label, category, and severity
  • Patients track mood from 1 to 5, add notes, and edit AI suggestions
  • Logs are stored per patient and visualized over time in My Health
  • Medication tracking supports adherence monitoring

Patients see clear trends. Clinicians receive structured, longitudinal data.

Clinical Note Cleaner (Clinician Side)

In the web app

  • Clinicians can view vitals, symptoms, medications and generate full SOAP notes in one click
  • Under the hood, we run a RAG pipeline that combines patient data and custom medical knowledge

Every assertion in the SOAP note is cited to:

  • Specific book sections including title, chapter, and page
  • Exact symptom log entries with ID and date

Citations stream inline with the note. Clinicians can save reports, get the change in the patient’s health status between visits called delta views.

End-to-end flow: voice or text input becomes structured patient data, which becomes a cited, structured SOAP note.

What We’re Proud Of

  • A full RAG and citation system built on medical books
  • Combining AI Health Diary and Clinical Note Cleaner into one coherent workflow
  • Clear temporal health visualizations
  • Strong Azure integration across Speech, OpenAI, and Functions
  • The clean and intuitive UI

What’s Next

  • Expand the medical knowledge base
  • Add authentication, access control, and audit logging
  • Implement per patient encryption and data separation

How We Built It

Mobile app: React, TypeScript, Tailwind, Azure Speech SDK
Web app: React with streaming SOAP and inline citations, Knowledge Base UI
Backend: Azure Functions in TypeScript, PostgreSQL with Prisma

SOAP generation uses:

  • Azure OpenAI
  • LangChain and LangGraph
  • A custom RAG pipeline over medical books using Azure AI Search
  • Streaming citations rendered live in the UI

We use synthetic and anonymized data only, based on a MIMIC-inspired schema . No real PHI used.

CareBridge AI transforms unstructured patient input into structured, cited, and trustworthy clinical intelligence across the care continuum.

Built With

Share this project:

Updates