《计算机应用》唯一官方网站 ›› 2025, Vol. 45 ›› Issue (6): 1784-1792.DOI: 10.11772/j.issn.1001-9081.2024060902

• 第十二届CCF大数据学术会议 • 上一篇    

基于图对比学习的再入院预测模型

姜超英1,2, 李倩1,2, 刘宁1,2(), 刘磊3, 崔立真1,2   

  1. 1.山东大学 软件学院,济南 250101
    2.山东大学 山东大学-南洋理工大学人工智能国际联合研究院,济南 250101
    3.山东省工业技术研究院,济南 250100
  • 收稿日期:2024-07-01 修回日期:2024-07-16 接受日期:2024-07-22 发布日期:2024-08-19 出版日期:2025-06-10
  • 通讯作者: 刘宁
  • 作者简介:姜超英(2001—),女,山东威海人,硕士研究生,CCF会员,主要研究方向:数据挖掘
    李倩(1993—),女,山东滨州人,助理研究员,博士,CCF会员,主要研究方向:知识图谱表示学习、知识推理、自然语言处理
    刘宁(1993—),男,山东日照人,助理研究员,博士,CCF会员,主要研究方向:医疗数据挖掘、知识增强 liun21cs@sdu.edu.cn
    刘磊(1981—),男,黑龙江哈尔滨人,教授,博士,CCF会员,主要研究方向:多媒体网络、软件定义网络
    崔立真(1976—),男,河北故城人,教授,博士,CCF会员,主要研究方向:大数据智能理论、数据挖掘、智慧科学和医疗健康、大数据AI。
  • 基金资助:
    山东省自然科学基金青年基金资助项目(ZR2022QF114)

Readmission prediction model based on graph contrastive learning

Chaoying JIANG1,2, Qian LI1,2, Ning LIU1,2(), Lei LIU3, Lizhen CUI1,2   

  1. 1.School of Software,Shandong University,Jinan Shandong 250101 China
    2.Joint SDU-NTU Centre for Artificial Intelligence Research,Shandong University,Jinan Shandong 250101,China
    3.Shandong Research Institute of Industrial Technology,Jinan Shandong 250100,China
  • Received:2024-07-01 Revised:2024-07-16 Accepted:2024-07-22 Online:2024-08-19 Published:2025-06-10
  • Contact: Ning LIU
  • About author:JIANG Chaoying, born in 2001, M. S. candidate. Her research interests include data mining.
    LI Qian, born in 1993, Ph. D., assistant research fellow. Her research interests include knowledge graph representation learning, knowledge reasoning, natural language processing.
    LIU Ning, born in 1993, Ph. D., assistant research fellow. His research interests include medical data mining, knowledge enhancement.
    LIU Lei, born in 1981, Ph. D., professor. His research interests include multimedia network, software defined network.
    CUI Lizhen, born in 1976, Ph. D., professor. His research interests include big data intelligence theory, data mining, smart science and medical health, big data AI.
  • Supported by:
    Youth Fund of Shandong Provincial Natural Science Foundation(ZR2022QF114)

摘要:

针对疾病间的共同作用与再入院情况的关系的挖掘不足以及相关模型泛化能力较弱的问题,提出一种基于图对比学习的再入院预测模型HealthGraph。首先,利用数据集中的疾病共现信息构建疾病编码图,以充分挖掘疾病之间的关联信息;其次,提出一种以图对比学习的思想为指导的患者数据增强方法,通过图采样器自适应地捕捉与任务相关的拓扑结构,构造新视图,提升数据丰富度,从而提高模型的泛化性能;最后,结合初始疾病编码图嵌入和新视图嵌入进行再入院预测。在真实数据集MIMIC-Ⅲ上构建呼吸系统疾病和循环系统疾病这2个数据集并进行大量实验。结果表明,相较于反转时间注意力模型(RETAIN)和阶段感知神经网络模型(StageNet),所提模型在准确率和F1指标上提升了1个百分点左右。此外,2组消融实验结果验证了所提模型在提高再入院预测的准确性和泛化性中的有效性。

关键词: 电子健康记录, 再入院预测, 图对比学习, 数据增强, 图神经网络

Abstract:

In order to solve the problems of the insufficient mining of relationship among inter-disease joint effects and readmission and the weak generalization ability of related models, a readmission prediction model based on graph contrastive learning was proposed, called HealthGraph. Firstly, the disease co-occurrence information in the dataset was used to construct a disease code map, so that the correlation information among diseases was fully explored. Then, a patient data augmentation method was proposed with the guidance of the idea of graph contrastive learning, and the topology related to the task was captured by the graph sampler adaptively, and a new view was constructed to improve the data richness, thereby improving generalization performance of the model. Finally, readmission prediction was carried out by combining the initial disease code map embedding and the new view embedding. The respiratory and circulatory system diseases datasets were constructed on real dataset MIMIC-Ⅲ and extensive experiments were conducted. The results show that compared with REverse Time AttentIoN model (RETAIN) and the Stage-aware neural Network model (StageNet), the proposed model has the accuracy and F1 indicators improved by about 1 percentage point. In addition, results of two groups of ablation experiments verify the effectiveness of the proposed model in improving the accuracy and generalization of readmission prediction.

Key words: Electronic Health Record (EHR), readmission prediction, graph contrastive learning, data augmentation, graph neural network

中图分类号: