🩺 Medical Report Analyzer

AI-Powered Blood Test Report Interpretation System

📌 Overview

Medical Report Analyzer is an AI-driven application designed to automate the extraction, interpretation, and summarization of blood test reports. The system leverages OCR, Large Language Models (LLMs), and an agent-based architecture to convert complex medical data into clear, patient-friendly insights.

Built with scalability and responsible AI principles in mind, this project demonstrates how multi-agent AI systems can be applied to real-world healthcare use cases.

---

🎯 Problem Statement

Blood test reports are often difficult for non-medical users to understand due to complex terminology and numerical reference ranges. Manual interpretation is time-consuming and requires medical expertise.

This project aims to:

• Automate blood report analysis
• Improve healthcare data accessibility
• Provide simplified, explainable insights for patients and students

---

🚀 Key Features

• 📄 Automated OCR Extraction from uploaded blood test reports
• 🤖 Agent-Based Architecture using CrewAI
• 🧠 LLM-Powered Medical Interpretation with reference range mapping
• 📝 Patient-Friendly Summaries generated via prompt engineering
• 🎛 Streamlit Web Interface for easy report upload and analysis
• 🔄 Modular & Scalable Design for future diagnostic extensions
• 🔐 Responsible AI Practices with no sensitive data retention

---

🏗 System Architecture

The system is built using a multi-agent pipeline, where each agent performs a specialized task:

1. OCR Parser Agent

   • Extracts text from uploaded blood reports using Tesseract OCR

2. Health Data Interpreter Agent

   • Identifies medical parameters
   • Maps values to standard reference ranges
   • Generates contextual medical insights using LLMs

3. Summary Generator Agent

   • Converts technical results into simplified, patient-friendly explanations

Agents are orchestrated sequentially using CrewAI to ensure clean data flow and modular execution.

---

🛠 Tech Stack

🔹 AI & NLP

• CrewAI (Agent orchestration)
• Hugging Face Transformers
• Prompt Engineering & Caching

🔹 OCR

• Tesseract OCR

🔹 Backend / Logic

• Python
• Modular agent workflows
• Structured data passing between agents

🔹 Frontend

• Streamlit (Interactive web UI)

---

📂 Project Structure

medical-report-analyzer/
│
├── agents/
│   ├── ocr_parser.py
│   ├── health_interpreter.py
│   └── summary_generator.py
│
├── app.py                  # Streamlit application
├── requirements.txt
├── README.md
└── utils/
    ├── prompts.py
    └── helpers.py

---

⚙️ Installation & Setup

1️⃣ Clone the Repository

git clone https://github.com/ahmasmibhanu/HealthLens.git
cd HealthLens

2️⃣ Install Dependencies

pip install -r requirements.txt

3️⃣ Run the Application

streamlit run app.py

---

🧪 Usage

1. Upload a blood test report (PDF or image)
2. OCR extracts the report text
3. AI agents analyze medical parameters
4. A simplified, patient-friendly summary is generated
5. Results are displayed in real-time on the UI

---

📊 Impact & Results

• Reduced manual effort in interpreting blood reports
• Improved healthcare data accessibility for non-technical users
• Achieved accurate extraction across diverse report formats
• Received positive usability feedback from medical students and early testers
• Demonstrated the feasibility of agent-based AI systems in healthcare

---

🔮 Future Enhancements

• Backend API using FastAPI
• Database integration for report history
• Explainability scores and confidence metrics
• Authentication and role-based access
• Support for additional diagnostics (X-ray, MRI reports)
• Deployment using Docker and cloud platforms

---

⚠️ Disclaimer

This tool is intended for educational and informational purposes only. It does not provide medical diagnosis or treatment advice. Always consult a qualified healthcare professional for medical decisions.

---

👨‍💻 Author

@ahmasmibhanu ~AI / Full Stack / Software Engineering Enthusiast

---

⭐ Acknowledgements

• CrewAI
• Hugging Face
• Streamlit
• Open-source OCR and NLP communities

---