Research On Decision Support For Cerebrovascular Diseases Based On Knowledge Graph

Cerebrovascular disease has the characteristics of high morbidity,high disability,and high mortality,and the patient’s economic burden is relatively high,which has attracted widespread attention from the society.In recent years,research on cerebrovascular diseases has achieved certain results,but the lifelong risk of stroke in my country is still the highest in the world at 39.3%,of which the lifelong risk of stroke in men is as high as41.1%.The prevention and treatment of stroke in my country is still facing huge challenges.With the development of medical informatization,hospitals have accumulated a large number of electronic medical records,which provide a data foundation for data mining and data analysis.At the same time,the development of big data technology provides good technical means for processing medical data,which is conducive to fully mining hidden information.Based on the above background,this paper analyzes the current research status of cerebrovascular disease,knowledge map and decision support system,and finds the necessity of assisting decision support research on cerebrovascular disease based on knowledge map.This paper uses electronic medical records as the data source to construct a knowledge map of cerebrovascular diseases,combined with knowledge representation learning,to achieve similar medical records retrieval,and to construct an auxiliary decision support system for cerebrovascular diseases.The main research contents of this article are as follows:(1)Knowledge graph research based on entity relationship extraction model: Aiming at the high cost of Chinese electronic medical record annotation,the lack of large-scale corpus,and the problem that a single vector of characters cannot represent ambiguity,the pre-training model BERT is used to vectorize the text.To more accurately represent the complex language structure and rich text semantic relationship,implement the BERT-BiLSTM-CRF model and conduct comparative experiments to improve the accuracy of entity and relationship recognition.Then build a knowledge map of cerebrovascular disease based on the entity relationship and store it in the map database.(2)Research on similar medical records retrieval based on knowledge graph and knowledge representation learning technology: Aiming at the problem of complex semantics and incomplete content of electronic medical records,the electronic medical records are represented based on knowledge graphs,and knowledge representation learning technology is used to vectorize knowledge graphs to realize knowledge representation The simulation and comparative analysis of the learning model obtain the vector representation of entities and relationships to form the text vector representation of the electronic medical record,and realize the calculation and sorting of the medical record similarity through the cosine similarity.(3)Research on assisted decision support for cerebrovascular diseases based on the knowledge map: design and implement an assisted decision support system for cerebrovascular diseases based on the knowledge map,and display the functional application interface,introduce the functions of the system in detail,and realize similar medical record retrieval and disease Diagnosis and treatment plan analysis,provide doctors with decision support in the diagnosis and treatment process.This paper first proves the effectiveness and accuracy of the BERT model in text vectorization through the comparative experiments of the four models of BERT-BiLSTM-CRF,BiLSTM-CRF,LSTM-CRF and CRF,and is based on the BERT-BiLSTM-CRF.And build a knowledge graph based on the entities and relationships extracted from the BERT-BiLSTM-CRF model.Secondly,through the comparative experiment of knowledge representation learning model,it proves the advantages of knowledge graph in realizing similar medical record retrieval.Finally,an auxiliary decision support system is realized to provide support for clinical diagnosis and treatment of cerebrovascular diseases.To promote prevention and control work.