Research And Application Of Medical Intelligent Question And Answer System Based On Deep Learning

With the rapid development of technology,artificial intelligence technology is widely used in various fields.In the field of medical services,traditional methods can no longer meet the needs of users,while the problem of unbalanced distribution of medical resources is becoming increasingly prominent,and intelligent question and answer systems are gradually becoming the service people prefer because of their convenience and efficiency.In recent years,with the continuous development of deep learning technology,deep learning methods are applied to medical intelligent Q&A systems to analyze and process medical data using deep learning methods to effectively solve the problems of Q&A matching,untimely manual responses and uneven distribution of medical resources,and improve the efficiency and quality of medical services.For this purpose,this paper researches and designs a medical intelligent question and answer system.This paper does the following on the research and application of medical intelligent question and answer system based on deep learning:(1)A medical entity recognition method based on semantic syntactic features and attention is improved.Based on the BERT pre-trained language model,this method enhances the feature representation and context coding ability of the input text through the improved bidirectional long short-term memory network and graph convolutional neural network.The attention mechanism is introduced to enrich the semantic feature representation of the sequence.The coding results are input into the decoding layer,and CRF is used to obtain the global optimal label sequence,so as to obtain the entities in the sentence.Ablation experiments show that the introduction of improved bidirectional GCN network coding and attention weight information can improve the accuracy of entity recognition tasks.(2)An improved medical automatic question answering method that integrates words to enhance semantic multi-attention is proposed.In the embedding layer,the fusion model of pre-training model BERT and WoBERT is used to extract and fuse the text vector representation based on word level and word level respectively to obtain more complete sentence vector semantic information.After that,the attention mechanism is added to generate the answer representation containing the question information,and input into the bidirectional gated recurrent unit to obtain the global semantic features of the sentence;finally,the multi-attention pooling module is used to characterize the interaction between the question and the answer,and the best answer is found by calculating the similarity of the question and answer.Experiments show that the vector representation of fused words and the addition of attention mechanism can improve the accuracy of the automatic question answering model.(3)A medical intelligent question answering system is designed and constructed.The medical entity recognition method based on semantic syntactic features and attention is used to realize the preliminary screening of medical problems,and the answers are found from the graph database according to the medical entity.For the problem that the medical entity is not extracted,the medical automatic question answering method that enhances semantic multi-attention by fusing words is used to realize the return of relevant answers through semantic understanding.Finally,the system is tested and analyzed.The experimental results show that the system has good performance.The improved medical entity recognition method and automatic question answering method in this paper are applied to the medical intelligent question answering system,which improves the function of the question answering system,improves the answer retrieval performance of the medical intelligent question answering system,and provides convenience for doctors and people seeking medical consultation.