Auxiliary Diagnosis Model Of Acute Myocardial Infarction Based On Knowledge Graph Embedding

Acute Myocardial Infarction(AMI)has many atypical symptoms,such as shoulder pain and neck pain,so that the clinical misdiagnosis rate is high.Artificial intelligence technology has been widely used in the medical field,but there is no research to improve the high misdiagnosis rate of acute myocardial infarction.Based on this research,on the one hand,it can improve the misdiagnosis of acute myocardial infarction.When patients with acute myocardial infarction choose other departments because of atypical clinical manifestations,such as shoulder pain and neck pain,the auxiliary diagnosis system can prompt doctors to check the patients with acute myocardial infarction in time.On the other hand,it can provide new ideas for text classification.Transfer learning based on pre-training model,fine-tuning skills,knowledge graph fusion and the skill of final classification can be applied to other text classification tasks.Generally speaking,this paper constructs a classification model of acute myocardial infarction.By inputting the patient’s medical record information into the model,the diagnosis result of the model can be obtained,whether the patient should be diagnosed as acute myocardial infarction.The result can be used as the basis of actual diagnosis.In this paper,877 electronic medical records are selected as training and testing data,of which 429 are medical records of patients with acute myocardial infarction.In the process of model construction,firstly,the entity vector is trained by TransD model based on knowledge graph,then fused with the text vector representation trained by Bert model,and finally classified by the fused vector.In this paper,when using Bert model,some fine-tuning operations are carried out,including the selection of output data and the setting of learning rate,in order to enhance the classification effect.Finally,the model achieved good results,with the accuracy rate of 0.9242,recall rate of 0.9538 and F1 value of 0.9254 on the test set.The test results show that the model has a strong ability to detect patients with acute myocardial infarction,taking into account the rational use of medical resources and the efficiency of diagnosis and treatment.The experimental results show that the fusion of knowledge graph and the two fine-tuning operations of Bert model have brought positive effects on improving the model effect.