计算机应用 ›› 2020, Vol. 40 ›› Issue (5): 1272-1277.DOI: 10.11772/j.issn.1001-9081.2019101766

• 人工智能 • 上一篇    下一篇

基于图卷积神经网络的医保欺诈检测算法

易东义, 邓根强, 董超雄, 祝苗苗, 吕周平, 朱岁松   

  1. 华中科技大学 协和深圳医院,广东 深圳 518060
  • 收稿日期:2019-10-18 修回日期:2019-12-12 出版日期:2020-05-10 发布日期:2020-05-15
  • 通讯作者: 邓根强(1991—)
  • 作者简介:易东义(1993—),男,湖北潜江人,硕士研究生,主要研究方向:深度学习、数据挖掘; 邓根强(1991—),男,湖南株洲人,硕士研究生,主要研究方向:模式识别、数据挖掘; 董超雄(1976—),男,湖北天门人,硕士研究生,主要研究方向:医院信息化建设、数据挖掘; 祝苗苗(1992—),女,湖南郴州人,硕士研究生,主要研究方向:智能算法、多目标优化; 吕周平(1985—),男,浙江慈溪人,主要研究方向:智能算法优化; 朱岁松(1964—),男,湖北咸宁人,主要研究方向:医保政策、医院信息化建设。
  • 基金资助:

    深圳市南山区技术研发和创意设计项目(深南科卫2018042号)。

Medical insurance fraud detection algorithm based on graph convolutional neural network

YI Dongyi, DENG Genqiang, DONG Chaoxiong, ZHU Miaomiao, LYU Zhouping, ZHU Suisong   

  1. Union Shenzhen Hospital,Huazhong University of Science and Technology, Shenzhen Guangdong 518060, China
  • Received:2019-10-18 Revised:2019-12-12 Online:2020-05-10 Published:2020-05-15
  • Contact: DENG Genqiang, born in 1991, M. S. candidate. His research interests include pattern recognition, data mining.
  • About author:YI Dongyi, born in 1993, M. S. candidate. His research interests include deep learning, data mining.DENG Genqiang, born in 1991, M. S. candidate. His research interests include pattern recognition, data mining.DONG Chaoxiong, born in 1976, M. S. candidate. His research interests include hospital informationization construction, data mining.ZHU Miaomiao, born in 1992, M. S. candidate. Her research interests include intelligent algorithm, multi-objective optimization.LYU Zhouping, born in 1985. His research interests include intelligent algorithm optimization.ZHU Suisong, born in 1964. His research interests include research on medical insurance policy, hospital informationization construction.
  • Supported by:

    This work is partially supported by Shenzhen Nanshan District Technology Research and Development and Creative Design Project (2018042).

摘要:

针对医疗保险欺诈检测当中欺诈样本不足、数据标注昂贵和传统基于欧氏空间的模型准确率低的问题,提出了一种新的基于图卷积和变分自编码的单分类医保欺诈检测模型(OCGVAE)。首先,通过病人就诊记录建立社交网络,计算病人和医生之间的权重关系,并设计了一个2层的图卷积神经网络(GCN)作为社交网络数据的输入,用以降低社交网络的数据维度;然后,设计了一个变分自编码(VAE)用以实现只存在一类欺诈样本标签的情况下的模型训练;最后,设计了一个逻辑回归(LR)模型用以判别数据类别。实验结果表明,OCGVAE模型的检测准确率达到87.26%,相较于一类对抗神经网络(OCAN)、一类高斯过程(OCGP)、一类近邻(OCNN)、一类支持向量机(OCSVM)和半监督图卷积神经网络(Semi-GCN)算法,分别高出16.1%、70.2%、31.7%、36.5%和27.6%,说明所提模型有效提高了医保欺诈筛查精度。

关键词: 医保欺诈检测, 图卷积神经网络, 变分自编码, 社交网络, 单分类, 主动学习

Abstract:

Aiming at the problems of insufficient fraud samples, expensive data labeling and low accuracy of traditional Euclidean space model, a new One-Class medical insurance fraud detection model based on Graph convolution and Variational Auto-Encoder (OCGVAE) was proposed. Firstly, a social network was established through patient visit records, the weight relationships between the patients and the doctors were calculated, and a 2-layer Graph Convolutional neural Network (GCN) was designed as the input of the social network data to reduce the data dimension of the social network. Secondly, a Variational Auto-Encoder (VAE) was designed to implement the model training under only one-class fraud sample label. Finally, a Logistic Regression (LR) model was designed to discriminate the data category. The experimental results show that the detection accuracy of the OCGVAE model reaches 87.26%, which is 16.1%,70.2%,31.7%,36.5%,and 27.6% higher than that of One-Class Adversarial Net (OCAN), One-Class Gaussian Process (OCGP), One-Class Nearest Neighbor (OCNN), One-Class Support Vector Machine (OCSVM) and Semi-supervised GCN (Semi-GCN) algorithm, demonstrating that the proposed model effectively improves the accuracy of medical insurance fraud screening.

Key words: medical insurance fraud detection, Graph Convolutional neural Network (GCN), Variational Auto-Encoder (VAE), social network, one-class, active learning

中图分类号: