Medical Entity Recognition Based on Bidirectional LSTM-CRF and Natural Language Processing Technology and Its Application in Intelligent Consultation

Knowledge graphs have become increasingly important in scientific research and technological applications, particularly in the medical field, making them a focal point in artificial intelligence research.With the rise of online medical communities, doctor-patient Q&A exchanges have emerged as valuable information sources.However, several challenges exist in processing this information: the technical nature of medical knowledge, unstructured text formats, and varied language expressions complicate both medical entity identification and relationship extraction.We developed an entity recognition framework that combines convolutional neural networks (CNN), bidirectional long short-term memory networks (BiLSTM), and conditional random fields (CRF).When tested on a breast cancer doctor-patient Q&A dataset, our framework achieved 92.32% recognition accuracy, outperforming competing models.Additionally, to address language expression variations, we incorporated BERT-Attention for relationship extraction, achieving an accuracy of 89.8%.Based on these results, we used Echarts to create visual representations of the medical knowledge graph and explored its applications in intelligent consultation systems.Our aim is to provide doctors with supportive diagnostic tools that can improve efficiency and contribute to the advancement of personalized medicine.