基于BP神经网络的健康保险欺诈识别研究
发布时间：2018-11-11 00:28
【摘要】：从全球范围来看,保险欺诈案件呈逐年递增态势,而且每年都会造成数千亿美元损失。保险欺诈不仅扭曲了保险定价机制,损害保险经营的最大诚信原则,而且还严重威胁医保基金安全,妨碍医保政策的有效实施。因此,反欺诈研究尤其是欺诈识别研究已成为学术界和实务界研究的热点领域。关于保险欺诈的理论研究已较为深入,研究人员从信息经济学、社会心理学等角度对欺诈的形式、成因及反欺诈措施等方面进行了系统研究；机动车保险欺诈的实证研究从最初的统计分析方法发展到人工智能识别技术以及两者的有机结合,并从概念模型构建向实证分析逐步过渡。然而,受健康保险信息技术和数据储备的限制,以及复杂医疗环境的阻碍,相对于国际上健康保险欺诈实证研究的热络,国内健康险欺诈识别与度量的研究还寥寥无几。 基于此,本文在参考国内机动车辆险欺诈识别研究成果的基础上,尝试运用统计回归和神经网络相结合的方法对健康服务需求方道德风险引致的欺诈进行识别研究。首先结合国内外保险欺诈的相关研究对我国健康险欺诈问题进行理论分析；以住院医疗保险具体险种为例搜集索赔案件样本数据,在理论分析的基础上结合专家意见确定欺诈识别因子；运用logistic回归分析提取具有模型显著性的识别因子,作为输入数据对构建的BP神经网络模型进行训练,并选取检验样本对模型的有效性进行预测检验；最后根据研究结果提出针对性的反欺诈措施和政策建议。研究结果显示,嵌入logistic回归分析的BP神经网络模型在特定条件下和一定范围内可以作为健康险欺诈识别的有效工具。本研究在一定程度上揭示了我国健康险欺诈的特征与规律,对改进保险机构欺诈识别技术、提高反欺诈能力具有一定的现实意义。
[Abstract]:Globally, insurance fraud cases are increasing year by year and cost hundreds of billions of dollars a year. Insurance fraud not only distorts the insurance pricing mechanism and damages the principle of the greatest integrity of insurance management, but also seriously threatens the security of medical insurance fund and hinders the effective implementation of medical insurance policy. Therefore, the research on anti-fraud, especially fraud identification, has become a hot area in academia and practice. The theoretical study on insurance fraud has been more in-depth, the researchers from the information economics, social psychology and other aspects of the form of fraud, causes and anti-fraud measures were systematically studied; The empirical study of motor vehicle insurance fraud developed from the initial statistical analysis method to artificial intelligence identification technology and the organic combination of the two, and gradually transition from conceptual model construction to empirical analysis. However, due to the limitation of information technology and data reserve of health insurance, and the hindrance of complex medical environment, there are few researches on identification and measurement of health insurance fraud in China compared with the international research on health insurance fraud. Based on this, this paper tries to use the method of combining statistical regression and neural network to identify the fraud caused by moral hazard on the demand side of health service based on the research results of domestic motor vehicle insurance fraud identification. Firstly, the theoretical analysis of health insurance fraud in China is made based on the related research of insurance fraud at home and abroad. Taking the inpatient medical insurance as an example to collect the sample data of the claim cases and determine the fraud identification factor on the basis of theoretical analysis combined with the expert opinion. The logistic regression analysis is used to extract the significant recognition factors of the model, and the BP neural network model is trained as input data, and the validity of the model is predicted by selecting the test samples. Finally, according to the results of the study, targeted anti-fraud measures and policy recommendations are put forward. The results show that the BP neural network model embedded in logistic regression analysis can be used as an effective tool for the identification of health insurance fraud under certain conditions and within a certain range. To a certain extent, this study reveals the characteristics and laws of health insurance fraud in China, which has a certain practical significance to improve the fraud identification technology of insurance institutions and improve the ability of anti-fraud.
【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP183;F842.684
[Abstract]:Globally, insurance fraud cases are increasing year by year and cost hundreds of billions of dollars a year. Insurance fraud not only distorts the insurance pricing mechanism and damages the principle of the greatest integrity of insurance management, but also seriously threatens the security of medical insurance fund and hinders the effective implementation of medical insurance policy. Therefore, the research on anti-fraud, especially fraud identification, has become a hot area in academia and practice. The theoretical study on insurance fraud has been more in-depth, the researchers from the information economics, social psychology and other aspects of the form of fraud, causes and anti-fraud measures were systematically studied; The empirical study of motor vehicle insurance fraud developed from the initial statistical analysis method to artificial intelligence identification technology and the organic combination of the two, and gradually transition from conceptual model construction to empirical analysis. However, due to the limitation of information technology and data reserve of health insurance, and the hindrance of complex medical environment, there are few researches on identification and measurement of health insurance fraud in China compared with the international research on health insurance fraud. Based on this, this paper tries to use the method of combining statistical regression and neural network to identify the fraud caused by moral hazard on the demand side of health service based on the research results of domestic motor vehicle insurance fraud identification. Firstly, the theoretical analysis of health insurance fraud in China is made based on the related research of insurance fraud at home and abroad. Taking the inpatient medical insurance as an example to collect the sample data of the claim cases and determine the fraud identification factor on the basis of theoretical analysis combined with the expert opinion. The logistic regression analysis is used to extract the significant recognition factors of the model, and the BP neural network model is trained as input data, and the validity of the model is predicted by selecting the test samples. Finally, according to the results of the study, targeted anti-fraud measures and policy recommendations are put forward. The results show that the BP neural network model embedded in logistic regression analysis can be used as an effective tool for the identification of health insurance fraud under certain conditions and within a certain range. To a certain extent, this study reveals the characteristics and laws of health insurance fraud in China, which has a certain practical significance to improve the fraud identification technology of insurance institutions and improve the ability of anti-fraud.
【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP183;F842.684
【参考文献】
相关期刊论文 前10条
1 王s，
本文编号：2324003
本文链接：https://www.wllwen.com/jingjilunwen/bxjjlw/2324003.html
