💊 MediScan AI

An AI-powered prescription reader that explains medicines in plain language — in your language.

---

Live Demo: https://mediscan-xh82u6gabtm94knuccriub.streamlit.app/

Overview

MediScan is a web application that helps patients understand their prescriptions. Doctors write prescriptions in complex medical shorthand that most patients cannot read. MediScan uses OCR and a large language AI model to extract the text, identify each medicine, and explain dosage, frequency, purpose, and warnings — in simple everyday language across 9 Indian languages.

Built in 24 hours for the UniMerge 1.0 Hackathon.

---

Problem

Millions of patients in India — particularly elderly individuals and those in rural areas — cannot interpret handwritten or printed prescriptions. Misreading dosage or frequency leads to preventable health complications. Access to a pharmacist who can explain is not always available.

Solution

Upload a photo of any prescription. MediScan reads it using OCR, sends the extracted text to an AI model, and returns a structured explanation with a plain-language summary. Results can also be read aloud for patients with low literacy or vision impairment.

---

Features

• Prescription image upload — JPG, PNG, JPEG support with OpenCV preprocessing for better OCR accuracy
• AI-powered analysis — Groq LLaMA 3.3 70B identifies medicines, dosage, frequency, purpose, duration, and warnings
• Editable OCR output — users can correct misread text before submitting to AI
• Manual text input — fallback for poor image quality
• 9 languages — English, Hindi, Odia, Marathi, Telugu, Punjabi, Tamil, Kannada, Bengali
• Read Aloud — full audio playback via gTTS
• Disclaimer — AI-generated output flagged clearly on every result

---

Tech Stack

Layer	Technology
Frontend	Streamlit + Custom CSS
AI Model	Groq API — LLaMA 3.3 70B Versatile
OCR	Tesseract OCR + OpenCV
Text-to-Speech	gTTS (Google Text-to-Speech)
Deployment	Streamlit Community Cloud
Language	Python 3.10+

---

Project Structure

mediscan-ai/
├── app.py                  ← Main Streamlit application
├── requirements.txt        ← Python dependencies
├── packages.txt            ← System packages (tesseract-ocr)
├── medicines.csv           ← Medicine reference data
└── .streamlit/
    └── secrets.toml        ← API key (not committed to repo)

---

Run Locally

1. Clone the repo

git clone https://github.com/anukta-acharya/mediscan-ai.git
cd mediscan-ai

2. Install Python dependencies

pip install -r requirements.txt

3. Install Tesseract OCR

• Windows → Download from github.com/UB-Mannheim/tesseract
• Linux → sudo apt install tesseract-ocr
• Mac → brew install tesseract

4. Add your Groq API key

Create .streamlit/secrets.toml:

GROQ_API_KEY = "gsk_your_key_here"

Get a free key at → console.groq.com

5. Run the app

streamlit run app.py

---

Deploy to Streamlit Cloud

1. Push this repo to GitHub
2. Go to share.streamlit.io → New app
3. Select your repo and app.py
4. Add GROQ_API_KEY under Settings → Secrets
5. Click Deploy

The packages.txt file handles Tesseract installation on the cloud automatically.

---

Dependencies

streamlit
groq
Pillow
pytesseract
opencv-python-headless
gTTS

---

Disclaimer

MediScan is for informational purposes only. Always consult a licensed doctor or pharmacist before taking any medication. Do not make medical decisions based solely on AI-generated output.

---

Made with dedication by Anukta Acharya  ·  UniMerge 1.0 Hackathon  ·  2026