Research On Profile Similarity Search And Disease Auxiliary Diagnosis Algorithms Based On Graph Sequence Model

With the explosive development of Internet technology,a large amount of data has emerged,and the data growth of the medical industry is particularly prominent.Medical big data is of great value,especially in clinically assisted care and health management.In the current Chinese medical service industry,the problem of imbalance between supply and demand has not been fully solved,and the medical resources and medical level in rural areas are more limited than in cities.The new smart healthcare model is a new healthcare service model that will rely on a new generation of user-friendly,real-time big data analytics and artificial intelligence and machine learning tools to provide healthcare service.Effective use of medical data and deep mining of data are the focus of future smart medical development.Medical data is often multimodal,including structured data and unstructured data.Medical data modeling has become a very important and challenging issue in medical big data analysis.The problems with most existing modeling methods are:ignoring multiple patterns of data,ignoring the temporal characteristics of medical records,and ignoring the explicit and implicit relationships between various medical features.We propose a knowledge graph and time series based approach to establish the connection between various types of multimodal data.And on this basis,we carry out research on profile graph similarity search and disease auxiliary diagnosis algorithms to support the efficient intelligent medical analysis task.1.First,a semantically rich medical knowledge graph database is built using medical dictionaries and actual clinical data collected from hospitals.2.We present a graph modeling approach to bridge the gap between different types of data,and fuse each patient's multimodal clinical data into a unified profile graph.On this basis,in order to capture the time evolution characteristics of a patient's clinical case,all of the patient's clinical cases are represented as an evolutionary sequence.3.We carry out research on profile graph similarity search and disease auxiliary diagnosis based on the graph model.A profile graph similarity search algorithm based on profile graph model is proposed.Based on this,we carry out research on a lazy learning disease auxiliary diagnosis algorithm based on graph sequence model.4.In order to evaluate the methods presented in this dissertation,the profile graph similarity search and disease auxiliary diagnosis experiments on ICU patients data and orthopedic patients data were performed.Experiments indicate that the research in this dissertation can effectively help doctors to understand the patient's medical information and provide useful information for the patient's disease auxiliary diagnosis.