A Knowledge Graph-Based Intelligent Q&A System for Rare Diseases

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Rare Disease Research

Background:

• Rare diseases present diagnostic and treatment challenges due to limited information.
• Existing medical knowledge retrieval systems often lack specialized capabilities for rare conditions.
• Integrating diverse data sources is crucial for comprehensive rare disease understanding.

Purpose of the Study:

• To develop and evaluate a knowledge graph-based intelligent question-answering (Q&A) system for rare diseases.
• To improve accessibility and retrieval of specialized medical knowledge for rare conditions.
• To leverage natural language processing (NLP) for efficient information extraction and intent recognition.

Main Methods:

• Construction of a knowledge graph integrating data on 126 rare diseases, 2,609 symptoms, and associated medical departments.
• Implementation of the Bert-BiLSTM-CRF model for named entity recognition (NER) achieving 82.13% accuracy.
• Utilization of the TextCNN model for intent recognition, reaching 94.54% accuracy.

Main Results:

• The developed system demonstrates effective entity extraction and intent recognition capabilities for rare disease queries.
• The knowledge graph successfully integrates a substantial volume of rare disease-related data.
• The system shows potential for enhancing user access to critical medical information concerning rare diseases.

Conclusions:

• The knowledge graph-based intelligent Q&A system offers a promising approach to addressing information gaps in rare disease care.
• The employed NLP models provide robust performance in understanding user queries within the rare disease domain.
• Further system optimization is recommended to enhance response times and broaden query handling capabilities for clinical application.