Prescription Analysis System

A cloud-native system for analyzing medical prescriptions using OCR, NER, and drug interaction checking. The system processes handwritten prescriptions and provides detailed analysis with safety warnings.

System Architecture

flowchart LR
    subgraph Cloud["GCP Cloud Platform"]
        GCS[("GCP Storage")]
        GKE["GKE Cluster"]
    end
    
    subgraph Processing["Prescription Analysis"]
        OCR["OCR Engine\nGPU Accelerated"]
        NER["NER Analysis"]
        DI["Drug Interaction\nChecker"]
        LLM["LLM Analysis"]
    end
    
    subgraph Output["Results"]
        Analysis["Analysis Report"]
        Warnings["Safety Warnings"]
        Instructions["Patient Instructions"]
    end
    
    GCS -->|Input Image| GKE
    GKE -->|Process| Processing
    OCR -->|Text| NER
    NER -->|Entities| DI
    DI -->|Drug Data| LLM
    LLM --> Output
    Output -->|Save| GCS

Loading

Features

• OCR Processing: GPU-accelerated text extraction from prescription images
• Named Entity Recognition: Identifies medications, dosages, and instructions
• Drug Interaction Analysis: Checks for potential drug interactions using OpenFDA
• Cloud Storage Integration: Works with GCP Storage for scalable processing
• Kubernetes Deployment: Ready for containerized deployment on GKE

Prerequisites

• Python 3.8+
• Docker
• Google Cloud SDK
• Kubernetes CLI (kubectl)
• OpenFDA API key

Quick Start

1. Clone and Setup:

git clone https://github.com/yourusername/prescription-analyzer
cd prescription-analyzer
make setup

2. Configure Environment:

# Create .env file
echo "OPENFDA_API_KEY=your_key_here" > .env
echo "GOOGLE_PROJECT_ID=your_project_id" >> .env

3. Run Locally:

make run

4. Deploy to GCP:

# Build and push Docker image
make docker-build
make docker-push

# Deploy to GKE
make deploy

Project Structure

├── src/
│   ├── ocr_processor.py      # OCR processing
│   ├── medication_ner.py     # Named entity recognition
│   ├── drug_interaction.py   # Drug interaction checking
│   └── prompt_llm.py        # LLM integration
├── kubernetes/              # Kubernetes configurations
├── scripts/                # Installation scripts
└── Makefile               # Build automation

API Usage

Process a prescription:

from src.main import process_prescription

results = process_prescription("path/to/image.jpg")
print(results['medications'])
print(results['warnings'])

Deployment

Deploy to GKE:

# Setup GKE cluster
gcloud container clusters create prescription-cluster \
    --region us-central1 \
    --num-nodes 3 \
    --machine-type n1-standard-4

# Deploy application
kubectl apply -f kubernetes/

Future Improvements

• Add support for multiple languages
• Implement real-time processing
• Add patient history integration
• Enhance drug interaction detection
• Add support for Azure and AWS

Email : [email protected]