ASD Detection
Transformer Architecture
What It Solves
Autism Spectrum Disorder screening using machine learning. Uses AQ-50 questionnaire (50 questions across 5 behavioral domains) with Transformer architecture for accurate screening. Includes demographic data processing, cross-platform hardware detection (GPU/CPU auto-selection), and clinical-grade Streamlit interface.
Comparison: Base Paper vs This Project
| Aspect | Base Paper | This Project |
|---|---|---|
| Architecture | Traditional ML (Logistic Regression) | Transformer (Self-Attention) |
| Questionnaire | AQ-10 (10 questions) | AQ-50 (50 questions) |
| Dataset | Single UCI dataset | 3 UCI datasets combined |
| Features | 10 features (AQ-10 only) | 19 features (AQ-50 + Demographics) |
| Model Params | N/A | ~170K parameters |
| Classification | Binary only | Multi-risk level (Low/Medium/High) |
| Interface | CLI | Professional Streamlit UI |
| Explainability | None | Feature Importance + XAI |
Tech Stack
Code Structure
asd-detection/
├── app.py # Streamlit web application
├── train.py # Model training script
├── test_pipeline.py # Testing pipeline
├── generate_report.py # Report generation
├── ablation_study.py # Ablation analysis
├── setup.py # Package setup
├── requirements.txt # Python dependencies
├── config/
│ └── settings.py # Configuration
│
├── src/
│ ├── data/
│ │ ├── aq50_questions.py # AQ-50 questionnaire (50 questions)
│ │ └── load_data.py # Data loading utilities
│ ├── preprocessing/
│ │ └── preprocessor.py # Data cleaning & encoding
│ ├── models/
│ │ ├── transformer.py # Transformer model
│ │ └── attention_cnn.py # CNN baseline
│ └── utils/
│ ├── helpers.py # Metrics, device detection
│ ├── clinical_utils.py # Clinical utilities
│ └── hardware_detect.py
│
├── tests/
│ └── test_asd.py # Unit tests
│
├── data/
│ └── diverse_features_processed.csv # Combined dataset
│
├── models/
│ ├── diverse_transformer.pt # Trained Transformer weights
│ └── diverse_model_package.joblib # Full model package
│
├── docs/
│ ├── README.md # Project documentation
│ ├── USER_GUIDE.md # User guide
│ └── TECHNICAL.md # Technical report What's Included
- Complete source code (Python/PyTorch)
- One documentation type FREE (IEEE Report / PPT / README)
- Additional documentation: +₹500 per type
- Original base papers + citation
- Trained model weights (.pt + .joblib)
- Demo videos
- Email support (7 days)
Base Paper
Get This Project
₹5,000 per project
Step 1: Share Your Requirements
Fill this form with your requirements, then message me on WhatsApp:
Fill Requirements FormStep 2: Message Me on WhatsApp
WhatsApp MeStep 3: Pay via UPI
Transfer ₹5,000 to:
8106968013@ibl
After payment, message me the UTR/RRN on WhatsApp
What's Included in Documentation
Each project comes with comprehensive documentation to help you understand and present your project.
Project Report
IEEE format documentation
Presentation Slides
PPT for viva presentation
README Guide
Step-by-step setup instructions
Base Paper
Original research paper cited
Note: These are the documents included. Click "View Details" on each project page to see actual project-specific documentation.