基于知识图谱表示学习哮喘预后预测模型  

Prognosis Prediction of Asthma Based on Knowledge Representation Learning of Medical Knowledge Graph

作　　者：刘兴惠 李至立 宋西成[2] 孙铭 LIU Xing-hui;LI Zhi-li;SONG Xi-cheng;SUN Ming(Shandong Vheng Data Technology Co.,Ltd,Yantai Shandong 264000,China;Department of Otolaryngology Head and Neck Surgery,Hospital of Yu Huang Ding,Yantai Shandong 264000,China)

机构地区：[1]山东纬横数据科技有限公司,山东烟台264000 [2]烟台毓璜顶医院,山东烟台264000

出　　处：《计算机仿真》2024年第11期517-522,共6页Computer Simulation

基　　金：中央引导地方科技发展专项资金(YDZX2021140);山东省重点研发计划(2020CXGC011302)。

摘　　要：为了实现哮喘患者居家监测正确的预后预测。结合医疗知识图谱、表示学习模型、神经网络方法构建了哮喘预后预测模型。模型首先分别构建哮喘患者和疾病知识图谱,引入表示学习模型对两类知识图谱进行向量映射,再将拼接后的患者和疾病向量矩阵通过卷积神经网络算法进行预后预测。仿真结果表明,获取的哮喘患者居家监测数据以及在已有多源数据基础上拓展纳入的全球哮喘倡议数据,有效缓解了哮喘患者和疾病知识图谱的数据稀疏问题;提出的TransR-CNN(TRC)模型对于哮喘疾病预后的预测效果较好,精度达到了94%,提高了计算效率和预测精度。To realize the correct prognosis prediction of home monitoring for asthma patients,combined with medi-cal knowledge graph,representation learning model and neural network method,the prediction model of asthma prog-nosis was constructed.Firstly the model constructs the knowledge graph of asthma patients and disease,and the vector map of the two knowledge graphs was carried out by introducing the representation learning model.Then,the patient and disease vector matrix after concatenation were predicted by the Convolutional Neural Network algorithm.Simula-tion results show that the home monitoring data of asthma patients obtained in this paper and the Global Initiative for Asthma(GINA)data expanded and included on the basis of existing multi-source data effectively alleviated the data sparseness of asthma patients and disease knowledge graph;the TransR-CNN(TRC)model proposed in this paper has a good predictive effect on the prognosis of asthma disease,with an accuracy of 94%,which improves the computation-al efficiency and prediction accuracy.

关 键 词：哮喘 知识图谱 表示学习:神经网络 预后预测 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]