Explore NeuroCypher ASD

May 30, 2025

NeuroCypher ASD is an AI-powered application that functions as both a screening assistant and an intelligent agent for detecting autism-related traits in toddlers. It combines graph-based knowledge representation, machine learning, and natural language processing to support research and educational efforts in early screening.

🌐 Access the app: http://195.251.143.63:8501

Key Features:

• ► Upload and analyze Q-Chat-10 and demographic screening data
• ► Constructed a complete and semantically rich knowledge graph representing all relevant entities and relationships through Neo4j
• ► Generate graph embeddings using Node2Vec
• ► Predict ASD likelihood using a trained XGBoost model
• ► Detect anomalies using Isolation Forest
• ► Identify similar cases through graph-based connections
• ► Ask natural language questions and receive Cypher-based answers using GPT4
• ► Visualize how traits and cases are interconnected in a knowledge graph

Why Used it?

• ► Acts as an intelligent agent powered by AI and graphs
• ► Implements a Graph-RAG (Retrieval-Augmented Generation) architecture
• ► Provides explainable, context-aware outputs
• ► Designed strictly for research and educational purposes

📝 Help us improve: Try the app and complete the short, anonymous usability survey (SUS):
https://forms.office.com/e/eiUuQkCDGa

👨‍💻 Developed by: Dr. Georgios Bouchouras, Dimitrios Doumanas, MSc and Assoc. Prof. K. Kotis. 🧪 Research Lab: Intelligent Systems Research Laboratory (i-Lab), University of the Aegean