Inspiration In many communities, access to timely medical attention is limited. We were inspired to create a solution that helps users triage their symptoms intelligently, reducing the burden on clinics and giving people clarity before visiting a doctor — especially during busy summer months. With conversational AI and a smart dashboard, we’ve fused healthcare, AI, and accessibility.

What It Does Triage AI is a web-based assistant that chats with users about their symptoms and assesses their condition. Based on their responses, it:

Gives them a triage color (🔴 red, 🟡 yellow, 🟢 green)

Suggests simple ways to reduce symptoms

Sends the summary to a secure doctor dashboard

Allows PDF export of the triage report

Offers reset & clear functions for privacy

(Coming soon) Language translation for non-English users

How We Built It Frontend: HTML, CSS, JavaScript (vanilla)

Backend: Node.js & Express

AI Brain: Groq API + OpenAI (GPT)

PDF Generation: PDFKit

Authentication: Custom session login for doctors

Local Testing: Ngrok for HTTPS tunneling

Hosted on: GitHub with plans for cloud deployment

Challenges We Ran Into Keeping the AI chat concise but helpful

Preventing prompt memory overload

Securing the doctor dashboard

Making sure files are downloaded and cleaned up properly

Aligning AI language with medical safety disclaimers

Accomplishments That We’re Proud Of Working AI triage chatbot with real-time feedback

PDF download of AI-generated summaries

Secure, session-protected doctor dashboard

End-to-end flow between user and doctor

What We Learned How to manage user input for prompt engineering

How to generate PDFs on demand and serve them securely

Balancing technical complexity with user simplicity

How to design for real-world medical sensitivity

What’s Next 🌍 Add language support (Spanish, Swahili, etc.)

📱 Make it mobile-first for community health workers

🩺 Build in Nurse/Patient roles

📊 Add analytics for triage case tracking

⚠️ Integrate SMS/USSD alerts in emergencies (e.g. via Africa's Talking)

Share this project:

Updates