Triage Assist

A web app designed to assist healthcare professionals in prioritizing patient care based on the severity of their condition using a Large Language Model.

Comment

Inspiration

We were inspired by the critical need for efficient and accurate triage, especially in high-pressure scenarios like mass casualty events or natural disasters. Traditional triage methods can be time-consuming and prone to human error, leading to delays in care and potentially life-threatening consequences. We wanted to create a solution that leverages cutting-edge AI technology to make triage faster, easier, and more accurate, ultimately saving lives and improving outcomes in emergency situations.

What it does

Triage Assist is an AI-powered tool that helps medical professionals and first responders quickly assess and prioritize patients based on their symptoms, vital signs, and other critical data. By inputting patient details such as age, pain level, blood pressure, heart rate, and oxygen saturation, the app generates a triage level (1–5) and provides a clear, color-coded visual representation of the patient's urgency. It also includes a description of the triage level to guide next steps. This ensures that patients receive the right care at the right time, even in chaotic environments.

How we built it

We built Triage Assist using the Groq API for AI-powered triage level generation and Streamlit for the user interface. The app takes user inputs (e.g., patient details, symptoms, and vital signs) and utilizes the LLaMA 3.3 LLM to process the data and return a triage level. We used Python for backend logic and integrated dynamic color coding and descriptions to make the results visually intuitive. The app is designed to be simple, fast, and accessible, even for users with minimal technical expertise.

Challenges we ran into

Developing Triage Assist presented several challenges. Handling incomplete or missing data inputs required robust error handling to ensure accurate results. Dynamic styling, especially updating colors and descriptions without caching issues, was tricky. Fine-tuning AI prompts for clinically accurate triage levels demanded extensive testing. Cloud hosting posed scalability and performance challenges, particularly for high-traffic emergency scenarios. Balancing simplicity with functionality was crucial to ensure usability in high-stress situations. Overcoming these hurdles made Triage Assist more reliable and user-friendly.

Accomplishments that we're proud of

We successfully integrated the Groq API with Streamlit to create a seamless, user-friendly experience. Implementing dynamic color coding and descriptions made triage levels visually intuitive and easy to understand. We built a robust tool that handles incomplete data gracefully and delivers accurate results. Most importantly, we created a scalable solution with the potential to significantly improve triage efficiency and accuracy in real-world emergency scenarios.

What we learned

Building Triage Assist taught us the importance of clear AI prompts for accurate outputs and robust error handling for incomplete data. We gained insights into dynamic UI updates and avoiding caching issues. Balancing simplicity with functionality was key for usability in high-stress scenarios. We also learned how to optimize cloud hosting for scalability and performance. Most importantly, we saw the potential of AI to transform emergency response, making processes faster, more accurate, and life-saving.

What's next for Triage Assist

In the future, we plan to expand Triage Assist by integrating with IoT devices, such as smart wearables, to automatically measure and input critical vitals like blood pressure, oxygen saturation, and heart rate. This will reduce manual data entry, improve accuracy, and speed up the triage process. We also aim to develop a mobile version for offline use in remote areas, add multi-language support for global accessibility, and incorporate advanced AI models to provide treatment recommendations. Additionally, a training mode will help medical students and first responders practice triage in simulated scenarios. By partnering with hospitals and emergency systems, we hope to make Triage Assist a standard tool for efficient and accurate triage in both routine and mass casualty events.

Built With

Share this project:

Updates