Knowledge Graph Construction And Question-Answering System Design For Dysphagia Care

China is entering an era of moderate aging,and elderly care has become a national strategy and a major livelihood issue that urgently needs to be addressed in today’s society.With the improvement of economic levels and medical conditions,the life expectancy of the elderly population continues to increase,and the proportion of people suffering from various age-related diseases is also on the rise.Among them,dysphagia is a high-incidence disease in the elderly population,with a morbidity rate as high as 66%.Dysphagia mainly manifests as a decrease in the patient’s chewing and swallowing abilities,which can trigger various complications such as aspiration pneumonia and malnutrition,seriously affecting the quality of life of the elderly and even threatening their safety.Caring for patients with dysphagia requires various professional care knowledge,which poses a challenge to many untrained elderly care providers.To address this issue,this project conducted knowledge collection and graph construction for dysphagia care,and developed a dysphagia care Q&A system based on this to meet the demand for professional knowledge during daily care.The research content of this article mainly includes the following three aspects:(1)Construction of a dysphagia care knowledge graph.In response to the fragmented and difficult-to-collect knowledge in the dysphagia care field,this article conducted knowledge collection and graph construction from three dimensions:graph ontology design,knowledge extraction,fusion,and storage.Firstly,based on the characteristics of the data,knowledge was extracted from several dysphagia literature and documents,using various methods such as character matching,regular expressions,and BERT to obtain entity and relationship information.Then,external knowledge graphs were used to expand the knowledge,and a dysphagia care knowledge graph was ultimately constructed.(2)Intention recognition and answer generation for dysphagia care Q&A.Based on the constructed knowledge graph,the second main task of this article is to identify intentions and generate answers for dysphagia care Q&A.According to the characteristics of dysphagia care Q&A,this article crawled text about dysphagia-related questions from a Chinese medical Q&A website and designed a self-training intention recognition model based on rules and active learning(STIR)to recognize user intentions.Based on this,a graph retrieval method combining graph traversal and template matching was used for answer queries.(3)Design and Implementation of a Question and Answer System for Swallowing Disorder Care.With the goal of providing professional and personalized knowledge on swallowing disorder care for non-professional home-based elderly care providers,and assisting them in providing professional care for swallowing disorders at home,this article designs a knowledge graph-based swallowing disorder care question and answer system.The system uses multiple mainstream technologies such as Neo4j graph database,natural language processing,and deep learning models to design the frontend,backend,database,and machine learning aspects.The main functional modules include a user-oriented question and answer module and an expert-oriented active learning annotation module.This work is the first attempt to construct a knowledge graph for swallowing disorder care and design a knowledge-based question and answer application.It embodies the development of a knowledge-driven elderly care model.In the future,we hope to digitize,process,and apply more professional knowledge on elderly care services,allowing elderly individuals and their caregivers to obtain professional care knowledge from home.