BioBrief

BioBrief turns confusing medical lab results into clear, plain-language explanations — in English or Twi — so every patient can understand their health, not just their doctor.

Comment

Inspiration

Every day, thousands of patients across Ghana and Africa receive printed lab results filled with medical jargon — HbA1c, eGFR, WBC differential — and walk home with no real understanding of what those numbers mean for their lives. Healthcare is expensive and inaccessible, but even when people do access it, the information they receive is locked behind terminology they were never taught. We built BioBrief to fix that

What it does

BioBrief is an AI-powered lab result interpreter. A patient takes a photo or uploads a scan of their printed lab result, and BioBrief:

  1. Reads the image using AI vision to extract every test name and value
  2. Explains each result in plain, friendly English — no medical degree required
  3. Translates explanations into Twi so Ghanaian patients can understand in their first language
  4. Suggests dietary and lifestyle changes to help improve borderline results
  5. Suggests nearby clinics to allow users seek more medical attention
  6. Has a symptom checker to allow users tell how they are feeling and how those symptoms relate to the results
  7. Recommends next steps and always advises the patient to consult a healthcare professional

How we built it

  • Frontend: React — a clean, mobile-first UI designed to feel warm and reassuring, not clinical
  • Backend: Node.js — handles API routing, image processing, and response formatting

Challenges we ran into

  • We had a little challenge picking the best models and one that would give us the best results.

Accomplishments that we're proud of

  • A working end-to-end prototype: upload a real lab result image and receive a plain-language breakdown within seconds
  • Twi language support: making this tool genuinely useful for a Ghanaian audience, not just a generic global product
  • A diet suggestion feature that connects lab values (e.g. high blood glucose) to practical, locally relevant food advice

What we learned

  • Prompt engineering for medical content requires extreme care — small wording changes dramatically affect the accuracy and tone of explanations

What's next for BioBrief

  • Expand language support to French, Hausa, and Pidgin English
  • Implement audio feature where translations are read to the user
  • The system should be able to take users' previous medical lab results and compare to the current one, and further tell the user the progress they have made with their condition.

Built With

Share this project:

Updates