CareLens

AI-powered video health consultation platform that helps users identify symptoms in real-time through live video and audio analysis.

Comment

Inspiration

Getting healthcare can be slow and stressful, especially when people don’t know where to start. We wanted to create something that feels like talking to a doctor — but faster and always available. That’s how CareLens was born: an AI-powered health check that listens, learns, and helps.

What it does

CareLens lets users talk to an AI through a live video call. The AI listens, asks questions, and watches for symptoms. As the conversation happens, it fills out a medical report and sends it to a doctor who can help — all automatically.

How we built it

We used:

  • Next.js + React for the frontend
  • Python (FastAPI) for the backend
  • OpenAI GPT-4 and Whisper for understanding speech and generating reports
  • WebRTC for live video We also designed the UI with Tailwind CSS and used Supabase for storage.

Challenges we ran into

  • Getting video, audio, and AI to work smoothly in real-time was tricky.
  • Whisper sometimes struggled with unclear audio.
  • Connecting all the parts (video, transcript, GPT-4, report) in a clean flow took a lot of debugging.

Accomplishments that we're proud of

  • We built a working AI video consultation flow from scratch.
  • The AI can now understand user speech and generate a medical summary.
  • We made it simple enough that anyone could use it — no tech background needed.

What we learned

  • How to work with speech recognition and video analysis in real time.
  • How to prompt GPT-4 to act like a doctor and ask follow-up questions.

What’s next for CareLens

  • Add doctor accounts so they can receive and respond to reports.
  • Let users upload their medical history and use that in future sessions.
  • Build a confidence/urgency score to tell users if they need urgent help.
  • Possibly work with clinics to test it out in the real world.

Built With

Share this project:

Updates