🩺 Medical Report Summarizer

A Flask-based web application that summarizes medical reports in simple, easy-to-understand language using LLaMA 3. Upload PDF files, extract relevant health information, and optionally translate the output into multiple languages.

---

📁 Project Structure

``` project-root/ │ ├── backend/ │ ├── app.py # Main Flask app
│ └── uploads/ # Directory for uploaded PDFs
│ ├── frontend/
│ ├── static/ # CSS/JS files
│ └── templates/
│ └── index.html # Frontend UI
│ ├── requirements.txt # Required Python libraries
└── README.md # This file
```

---

🚀 Features

• 📄 Upload PDF medical reports
• 🧠 Extract:
  • Patient Info
  • Diagnosis Overview
  • Medical History
  • Prescription Summary
  • Treatment Plan
  • Lifestyle Recommendations
  • Overall Summary
• 🌍 Multilingual support
• 💬 LLaMA 3 powered backend
• 🖥️ User-friendly interface

---

🛠️ Tech Stack

• Frontend: HTML, CSS, JavaScript, Jinja2
• Backend: Python, Flask
• LLM: LLaMA 3 via Ollama
• PDF Parser: PyPDF2
• Markdown Rendering: markdown2

---

🧪 Get Started With Ollama

---

1. Make sure your Ollama is installed properly, for that go to the website

https://ollama.com/download

2. Setup Ollama application on your local machine

3. To Install a model onto your system

Ollama #to check if ollama is properly installed

ollama run llama3 

---

🧪 Installation

---

1. Clone the repository

git clone https://github.com/your-username/medical-report-summarizer.git
cd medical-report-summarizer

2. Create a virtual environment

python -m venv venv
source venv/bin/activate  # On Windows use: venv\\Scripts\\activate

3. Install dependencies

pip install -r requirements.txt

4. Run the app

cd backend
python app.py

Visit `http://127.0.0.1:5000\` in your browser.

📊 Model Evaluation (Sample)

Metric	Score
ROUGE-L	0.1943
BLEU	0.2616
BERTScore F1	0.8837
BERT Precision	0.9035
BERT Recall	0.8648

---

📌 Notes

• Ensure Ollama and the LLaMA 3 model are set up correctly.
• The model may need an internet connection or local cache for best performance.
• Translation quality may vary by language and report structure.

---