基于BERT-CNN模型的病案数据智能检索与分类研究  

Research on intelligent retrieval and classification of medical record data based on BERT⁃CNN model

作　　者：高洁 王琳[1] GAO Jie;WANG Lin(Tangshan Maternal and Child Health Hospital,Tangshan 063000,China)

机构地区：[1]河北省唐山市妇幼保健院,河北唐山063000

出　　处：《电子设计工程》2025年第9期27-30,37,共5页Electronic Design Engineering

摘　　要：针对病案数据量大、结构复杂以及检索与分类困难等问题,提出了一种基于深度学习模型的智能检索与分类方法。通过使用BERT模型对病案文本数据进行词向量转换,将这些向量输入CNN模型,实现局部特征提取,并采用全连接层对病案数据进行分类。实验结果表明,该方法在分类任务中展现出较高的精度。BERT与CNN的结合不仅增强了病案数据的语义表征能力,还提升了分类效果,加权精度为87.22%,加权召回率为85.30%。说明该方法为医疗信息处理提供了有效支持。A deep learning model-based intelligent retrieval and classification method is proposed to address the problems of large volume,complex structure,and difficulty in retrieval and classification of medical record data.By using the BERT model to perform word vector transformation on medical record text data.Input these vectors into the CNN model to achieve local feature extraction,and use a fully connected layer to classify medical record data.The experimental results show that this method exhibits high accuracy in classification tasks.The combination of BERT and CNN not only enhances the semantic representation ability of medical record data,but also improves the classification performance,weighted accuracy is 87.22%,weighted recall rate is 85.30%.This method provides effective support for medical information processing.

关 键 词：病案数据 深度学习 词向量 特征提取 分类 

分 类 号：TN99[电子电信—信号与信息处理]