Research On The Construction Technology Of Disease Knowledge Graph For Diagnosis Report

With the economic development,people's life rhythm continues to accelerate,and health problems are becoming increasingly prominent.Smart medicine meets people's needs and future development trends in the medical field.With the development of algorithms such as deep learning and the wide application of artificial intelligence in various fields.The application of artificial intelligence in the medical field to develop smart medicine can facilitate data sharing and in-depth use of data,make better use of medical resources,and simplify medical procedures.The knowledge graph is the basis for building intelligent medical care and can provide a data basis for efficient medical services.A large number of medical images and corresponding diagnostic reports in the traditional medical field have high excavation value,and diagnostic reports are an important basis for doctors' diagnosis and treatment.A complete and usable diagnostic report contains medical images and inferences about the patient's illnesses as seen and obtained by the doctor's professionally described images.This article uses the diagnosis report collected from the Department of Orthopedics of Kunming Traditional Chinese Medicine Hospital as the data base to study and construct an orthopedics consultation platform based on knowledge graph.The main contents include:First,extract the entity relationship in the diagnosis report to obtain the basic data "entity-relationship-entity" triple that constructs the knowledge graph.According to the characteristics of the diagnostic report,the labeling strategy developed under the guidance of relevant hospital experts first performs a series of processing and labeling on the diagnostic report,and uses the deep learning model based on the two-way longshort memory network to train and learn to extract relatively accurate entity relationships.The high noise caused by the equipment during medical imaging and the edge blur caused by the insignificant difference between organs and tissues optimize the image preprocessing,and use the difference superposition based on the center point to strengthen the edge difference and take the median gray value to Eliminate scattered noise points in the area.The effectiveness of the method is proved through experiments.Second,the current knowledge graph construction only uses the triple structure data extracted from the text information,while ignoring the large amount of text semantic description information,medical images and numerical information that exist in reality.In this paper,the triples and other information ignored in the diagnostic report are embedded into the low-dimensional dense vector space.It can make the triplets obtained from the diagnosis report higher in quality and contain more information,and to some extent realize knowledge graph complementation.Then for the problem that there is more than one disease name in a disease description,the method of clustering according to semantic similarity is used for knowledge fusion,and a good accuracy rate is obtained.