SafeMeds

📖 Overview

SafeMeds is a real-time, AI-powered clinical decision support tool that evaluates drug-to-drug interactions using natural language queries.
It helps healthcare professionals reduce medication risks and make faster, safer decisions.

🚀 Inspiration

Every year, thousands of patients experience adverse drug events due to complex or overlooked drug interactions. We noticed a gap in intuitive, real-time tools that help healthcare providers assess these risks efficiently. Our inspiration was to leverage the power of modern LLMs to make this process faster, more accessible, and more reliable for medical professionals.

💡 What it does

SafeMeds allows users—primarily healthcare providers—to enter natural language queries like
"Can I take ibuprofen with metformin?",
and receive immediate, AI-powered feedback about potential drug interactions and safety concerns.

Key capabilities:

Interprets complex drug combinations using natural language
Provides concise, medically-informed risk assessments
Offers real-time, responsive results to assist in decision making

🧱 How we built it

Frontend:

React.js
Deployed on Vercel
Clean and responsive UI tailored for medical environments

Backend:

Python Flask API
Hosted on AWS EC2 and Render for redundancy and performance
Connects securely to the frontend and LLM API

AI Model:

LLaMA3-8B via Groq API
Utilized for fast and contextually rich interpretation of drug interaction queries

Deployment:

Frontend on Vercel with custom domain
Backend deployed via AWS EC2 and Render
Full-stack integration with RESTful API endpoints

🧗‍♂️ Challenges we ran into

Latency: Ensuring real-time responses while working with large language models required careful infrastructure optimization.
Medical accuracy: Interpreting medical data accurately via an LLM required prompt engineering and validation loops.
Deployment hurdles: Coordinating between Vercel (frontend) and AWS EC2 (backend) involved SSL and CORS configuration challenges.

🏆 Accomplishments that we're proud of

Seamless integration of a powerful LLM (LLaMA3 via Groq) with a real-time web interface
Building a clean, responsive UI that's both fast and user-friendly
Deploying a fully operational, scalable full-stack app with live AI inference