Medical Worker Portal with AI X-ray Analysis

A comprehensive web-based medical portal that allows healthcare workers (doctors, nurses, radiologists) to access patient information and perform AI-assisted X-ray analysis with comparison capabilities.

Features

Authentication & Access Control

• Secure login system for medical workers
• Role-based access control (Doctor, Nurse, Radiologist)
• Doctors can only access their assigned patients
• Session management

Patient Management

• 100 pre-generated fake patient records
• Comprehensive patient demographics
• Detailed medical history including:
  • Allergies
  • Chronic conditions
  • Current medications
  • Previous surgeries
  • Family medical history
  • Blood type

AI-Powered X-ray Analysis

• HOPPR AI Integration: Analyzes chest X-rays using state-of-the-art radiology AI
• Doctor Input: Doctors provide their own diagnosis and findings
• Gemini AI Comparison: Automatically compares doctor's assessment with AI results
• Agreement Analysis: Provides detailed comparison showing:
  • Agreement level (High/Medium/Low)
  • Key agreements and differences
  • Clinical significance
  • Recommendations for follow-up

Supported X-ray Conditions

The system includes sample DICOM images for:

• Aortic enlargement
• Atelectasis
• Calcification
• Cardiomegaly
• Consolidation
• Interstitial Lung Disease (ILD)
• Infiltration
• Lung Opacity
• Normal studies
• Pleural effusion
• Pleural thickening
• Pneumothorax
• Pulmonary fibrosis

Installation

Prerequisites

• Python 3.8 or higher
• pip package manager

Setup Instructions

1. Navigate to the project directory:

   cd c:\Users\nours\Downloads\hackathon

2. Install required packages:

   pip install -r requirements.txt

3. Generate fake patient data:

   python generate_fake_data.py

   This will create:

   • 20 doctors
   • 10 nurses
   • 5 radiologists
   • 100 patients with complete medical histories

4. Start the web application:

   python app.py

5. Access the portal: Open your web browser and navigate to:

   http://localhost:5000
   

Usage

Login Credentials

All accounts use the password: password123

Sample Doctor Logins:

• Username: drsmith1, drjohnson1, drwilliams1, etc.

Sample Nurse Logins:

• Username: nursebrown1, nursejones1, etc.

Sample Radiologist Logins:

• Username: radgarcia1, radmiller1, etc.

Performing X-ray Analysis

1. Login to the portal using your credentials
2. Select a patient from the dashboard
3. Click "New X-ray Analysis" button
4. Enter your assessment:
   • Diagnosis
   • Clinical findings
   • Severity (mild/moderate/severe)
5. Select a DICOM file from the available X-rays
6. Click "Analyze with AI"
7. Review the comparison between your assessment and the AI analysis

Understanding Results

The results page shows:

• Your Assessment: Your diagnosis, findings, and severity rating
• AI Analysis: HOPPR AI's diagnosis, findings, and confidence level
• Gemini Comparison: Detailed comparison including:
  • Agreement percentage
  • Points of agreement
  • Key differences
  • Clinical significance
  • Recommendations

Project Structure

hackathon/
├── app.py                          # Main Flask application
├── generate_fake_data.py           # Data generation script
├── hoppr_integration.py            # HOPPR AI integration
├── gemini_comparison.py            # Gemini AI comparison logic
├── requirements.txt                # Python dependencies
├── templates/                      # HTML templates
│   ├── base.html                   # Base template
│   ├── login.html                  # Login page
│   ├── dashboard.html              # Main dashboard
│   ├── patient_detail.html         # Patient information
│   ├── analyze_xray.html           # X-ray analysis form
│   └── analysis_result.html        # Analysis results
├── HackathonSampleDICOMImages/     # DICOM image files
└── medical_portal.db               # SQLite database (created after setup)

Database Schema

MedicalWorker

• Login credentials
• Full name, role, department
• Relationship to patients

Patient

• Demographics (age, sex, blood type)
• Medical history
• Assigned doctor

XRayAnalysis

• Doctor's assessment
• AI analysis results
• Gemini comparison data
• Timestamps and file references

API Keys

The application uses the following APIs:

• HOPPR AI: For chest X-ray analysis
• Gemini AI: For comparing doctor and AI assessments

API keys are configured in:

• classification_demo_single.py: Contains both API keys
• hoppr_integration.py: HOPPR API key
• gemini_comparison.py: Gemini API key

Security Notes

• This is a demonstration/educational project
• Passwords are simplified for demo purposes
• In production, use proper password hashing and environmental variables
• Implement proper HIPAA compliance for real medical data
• Use HTTPS for all communications
• Store API keys securely

Technologies Used

• Backend: Flask, SQLAlchemy
• Frontend: Bootstrap 5, Bootstrap Icons
• AI/ML:
  • HOPPR AI (Medical imaging analysis)
  • Google Gemini (Natural language comparison)
• Medical Imaging: pydicom for DICOM file handling
• Database: SQLite

Demo Workflow Example

1. Login as drsmith1 / password123
2. View your assigned patients on the dashboard
3. Click on a patient (e.g., "John Doe")
4. Review patient's medical history
5. Click "New X-ray Analysis"
6. Enter your diagnosis:
   • Diagnosis: "Pleural effusion, moderate"
   • Findings: "Blunting of right costophrenic angle with opacity in lower right lung field"
   • Severity: "moderate"
7. Select a DICOM file from the Pleural_effusion folder
8. Click "Analyze with AI"
9. Review the detailed comparison showing agreement/differences

Troubleshooting

Database not found:

python generate_fake_data.py

Port 5000 already in use: Edit app.py and change:

app.run(debug=True, host='0.0.0.0', port=5001)

Missing dependencies:

pip install -r requirements.txt --force-reinstall

Future Enhancements

• Real-time DICOM viewer
• Export analysis reports to PDF
• Multi-factor authentication
• Advanced search and filtering
• Audit logging
• Integration with hospital information systems
• Support for additional imaging modalities (CT, MRI)

License

This is an educational/demonstration project for the hackathon.

Contact

For questions or issues, please refer to the project documentation.