OCD Detector

An AI-powered system for detecting and analyzing OCD patterns in digital communications

Overview

OCD Detector is a cutting-edge artificial intelligence system designed to identify and analyze Obsessive-Compulsive Disorder (OCD) patterns across multiple communication channels. By leveraging advanced machine learning technologies, our system provides valuable insights for mental health professionals and researchers.

The National Institute of Mental Health (NIMH) research indicates that individuals with OCD exhibit distinct patterns in their communication and behavioral manifestations[²], including:

• Repetitive thoughts and behaviors
• Specific language patterns and word choices
• Time-consuming rituals that interfere with daily activities
• Persistent intrusive thoughts

Our system aligns with the NIH's dimensional approach to OCD assessment[³], analyzing these patterns across digital communications to support clinical evaluation and research.

Key Technologies

• 🧠 Advanced Natural Language Processing (NLP)
• 👁️ Optical Character Recognition (OCR)
• 🎤 Voice Recognition and Analysis
• 🤖 Integration with OpenAI's GPT models
• 📱 CoreML Support for Apple Devices

Methodology

Our OCD detection system employs sophisticated algorithms and statistical methods to analyze behavioral patterns. See Research for more details.

The core methodology includes:

Pattern Recognition

We utilize a modified attention mechanism that weighs behavioral markers ($b_i$) against temporal frequency ($f_t$):

$$ A(b, f) = \sum_{i=1}^{n} \frac{b_i \cdot f_t}{\sqrt{d_k}} $$

where $d_k$ represents the dimensionality of the behavioral space.

Severity Scoring

The severity score ($S$) is calculated using a weighted combination of detected patterns:

$$ S = \alpha \sum_{i=1}^{n} w_i p_i + \beta \log(\frac{f_{obs}}{f_{exp}}) $$

where:

• $w_i$ represents pattern weights
• $p_i$ represents pattern intensities
• $f_{obs}$ is observed frequency
• $f_{exp}$ is expected frequency
• $\alpha, \beta$ are tuning parameters

Confidence Estimation

Our confidence metric ($C$) incorporates both model uncertainty and data quality:

$$ C = \left(1 - \frac{\sigma_m^2}{\sigma_{max}^2}\right) \cdot \left(\frac{n_{valid}}{n_{total}}\right) $$

where $\sigma_m^2$ represents model variance and $n_{valid}$ represents the number of valid data points.

Research Used

This project is based on published scientific research papers, please refer to the Research for more details.

Features

Multi-Modal Analysis

• Text Analysis

  • Real-time chat monitoring
  • Historical conversation analysis
  • Pattern recognition in messaging
  • Sentiment analysis

• Image Processing

  • Screenshot analysis
  • Text extraction from images
  • Visual pattern recognition
  • Metadata analysis

• Voice Recognition

  • Speech-to-text conversion
  • Vocal pattern analysis
  • Tone and rhythm detection
  • Real-time monitoring

AI Integration

• Custom fine-tuned models for OCD detection
• Integration with OpenAI's ChatGPT
• Expandable model architecture
• Transfer learning capabilities

Deployment Options

• Cloud-based deployment
• On-premise installation
• Mobile deployment via CoreML
• Edge device support

Quick Start

Prerequisites

• Python 3.8+
• Node.js 16+
• Tesseract OCR
• GPU recommended for training

Basic Installation

# Clone the repository
git clone https://github.com/skytells-research/ocd-detector.git

# Install dependencies
cd ocd-detector
pip install -r requirements.txt

# Start the application
python run.py

For detailed installation instructions, see Installation Guide.

Documentation

• Installation Guide
• Training Guide
• API Reference
• Model Architecture
• Contributing Guidelines
• Code of Conduct

Use Cases

Clinical Research

• Pattern identification in patient communications
• Long-term behavior tracking
• Treatment effectiveness monitoring

Mental Health Support

• Early detection of OCD tendencies
• Progress monitoring
• Intervention timing optimization

Personal Monitoring

• Self-awareness tools
• Progress tracking
• Pattern identification

Model Performance

Analysis Type	Accuracy	Precision	Recall
Text	89%	87%	91%
Image	76%	79%	74%
Voice	82%	84%	81%

Contributing

We welcome contributions from the community! Please read our Contributing Guidelines before submitting pull requests.

Development Setup

# Create virtual environment
python -m venv venv
source venv/bin/activate

# Install development dependencies
pip install -r requirements-dev.txt

# Run tests
pytest

Running the Application

Running the Frontend

To start the frontend application on port 8502:

streamlit run frontend/app.py --server.port 8502

Running the Backend

Please refer to the Running the Backend for detailed instructions.

Citation

If you use this project in your research, please cite:

@software{ocd_detector2024,
  author = {Skytells Research},
  title = {OCD Detector: AI-Powered OCD Pattern Analysis},
  year = {2024},
  url = {https://github.com/skytells-research/ocd-detector}
}

Support

• 📫 Report Issues
• 💬 Community Discussions
• 📖 Documentation
• 🤝 Contributing

Contributors

• @skytells-research
• @DrHazemAli
• @jenniferdou

Sponsor The Project

If you find this project useful, consider sponsoring us on GitHub to support ongoing development and maintenance. Sponsor Us

Acknowledgments

• OpenAI for GPT integration support
• TensorFlow team for model architecture insights
• Our amazing contributors and research partners

References

[¹] World Health Organization. (2022). Obsessive-compulsive disorder fact sheet. WHO Mental Health Atlas.
https://www.who.int/news-room/fact-sheets/detail/obsessive-compulsive-disorder

[²] National Institute of Mental Health. (2023). Obsessive-Compulsive Disorder: Signs, Symptoms and Treatment.
https://www.nimh.nih.gov/health/topics/obsessive-compulsive-disorder-ocd

[³] Mataix-Cols, D., et al. (2016). Dimensional Assessment of Obsessive-Compulsive Disorder: National Institute of Mental Health Proceedings. NIH Public Access.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4332094/

License

This project is licensed under the MIT License - see the LICENSE file for details.

For more info, Please contact Skytells Research on our website