Back to Projects
AI/ML Healthcare

ASD Detection

Transformer Architecture

100%
Accuracy
Multi-source dataset (Adult + Children + Adolescent)
Dataset
Latest
Tech (2026)
₹5,000
Price

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

PyTorchStreamlitTransformersscikit-learnReportLab

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

ASD Detection Using Deep Learning with AQ-50 Questionnaire

arXiv, 2024

View 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 Form

Step 2: Message Me on WhatsApp

WhatsApp Me

Step 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.