AI drug Identification system

AI-Based Drug Detection & Safety Analysis.

Comment

Inspiration

Medication misuse and incorrect drug consumption are major healthcare problems worldwide. Many people consume medicines without understanding their dosage, ingredients, or safety warnings. In emergency situations, identifying unknown pills quickly can be difficult and may lead to serious medical risks. We were inspired to create an intelligent AI-powered system that can instantly identify drugs and supplements using images and provide accurate medical information in a simple and accessible way.

What it does

MediScan AI is an AI-based drug identification and supplement analysis system that helps users identify medicines by uploading or capturing pill images. The system uses Artificial Intelligence, OCR, and live medical APIs to analyze the medicine and display important information such as:

Drug name Active ingredients Dosage details Safety warnings Contraindications Manufacturer information Confidence score of prediction

The platform also stores user scan history and integrates with live FDA medical databases for real-time drug information.

How we built it

We built the system using modern AI and web technologies.

Frontend React.js HTML/CSS JavaScript Backend FastAPI (Python) Database PostgreSQL AI & Machine Learning OpenCV for image preprocessing OCR (Tesseract) for text extraction YOLO / CNN-based model for pill identification APIs openFDA API RxNorm API DailyMed API Workflow User uploads or captures a medicine image Image preprocessing removes noise and isolates the pill OCR extracts visible text from the medicine AI model predicts the drug identity APIs fetch live drug details Results are displayed with warnings and dosage information

Challenges we ran into

During development, we faced several technical and practical challenges:

Identifying pills with blurry or low-quality images Handling medicines with similar shapes and colors Finding free and reliable medical APIs Understanding FDA API documentation and endpoints Integrating OCR with machine learning models Managing real-time API responses and caching Handling unsupported or missing drug data Designing a fast and user-friendly interface

We also learned that some medicine brand names are region-specific and may not exist in FDA databases.

Accomplishments that we're proud of

We are proud that we successfully created:

✅ An AI-powered medicine identification system ✅ OCR-based drug text extraction ✅ Real-time medical API integration ✅ Automatic dosage and warning display ✅ Secure database storage for scan history ✅ FastAPI backend with scalable architecture ✅ A healthcare-focused AI application with real-world impact

The project demonstrates how AI can improve healthcare accessibility and medicine awareness for everyday users.

What we learned

Through this project, we learned:

How AI and computer vision work in healthcare applications Integration of live medical APIs such as openFDA and RxNorm OCR and image preprocessing techniques FastAPI backend development Database design using PostgreSQL Handling real-world medical data and API limitations Importance of healthcare safety disclaimers and ethical AI usage

We also improved our problem-solving, debugging, and system design skills.

What's next for AI Drug Identification System

In the future, we plan to add:

Barcode and QR code scanning Mobile application support Voice assistant integration Multi-language support AI chatbot for medicine guidance Prescription scanning and analysis Drug interaction checker Counterfeit medicine detection Offline AI model support Cloud deployment and scalability improvements

Built With

Share this project:

Updates

Submission history