Quick Triage

Inspiration

"Quick Triage" was inspired by the need to improve emergency response systems. Recognizing that every second counts in medical emergencies, the project uses AI to streamline the triage process, helping dispatchers quickly assess call severity and allocate resources efficiently, reducing delays and human error.

What it does

Quick Triage transcribes an emergency call and then uses an LLM to extrapolate critical caller data, such as their name, location, emergency type, emergency status, and the recommended branch and medical recommendation of emergency services to be alerted.

How we built it

Backend: Python We used Python for the backend to handle machine learning tasks, data processing, and AI model integration. Frontend: Next.js The frontend was built with Next.js, providing a fast, responsive user interface with server-side rendering for real-time updates. Fast APIs: • Llama3-8b-8192 (Text Model): This model processes text input from emergency calls, categorizing symptoms and providing triage recommendations. • OpenAI Whisper (Voice Model): Whisper transcribes voice calls into text, enabling real-time speech-to-text conversion for better accessibility and understanding. Deployment: Vercel We deployed the project on Vercel, ensuring seamless integration of frontend and backend with scalable, low-latency performance.

Challenges we ran into

We faced challenges with data accuracy, ensuring real-time performance, and overcoming skepticism from users. Implementing AI features was difficult due to our limited experience, and we struggled with frontend-backend integration after transitioning to a new framework.

Accomplishments that we're proud of

We successfully implemented AI for speech-to-text transcription and used an LLM to extract critical emergency data, streamlining the dispatch process. And AI integration, improved decision-making, and a user-friendly interface were key milestones, supported by collaboration with experts.

What we learned

We gained new frontend and backend development skills, learned how to integrate various AI technologies, and developed time management and project planning skills under tight deadlines. AI can be effectively applied to improve decision-making in high-pressure environments.

What's next for Quick Triage

• We plan to refine models to improve accuracy and handle more diverse situations. • Develop mobile versions and add multi-language capabilities for broader reach. • Real-Time Analytics and Reporting

Built With

• fastapi
• llama-8b
• llama3-8b-8192
• nextjs
• openai-whisper
• python
• vercel