基于BP神经网络模型的我国城乡居民基本医疗保险运行风险预警及防范策略研究

The Integration of the basic medical insurance for urban residents and the NCMS is one of the trend of health reform in China, which thought to improve the inequity of system, and more important to ensure the medical insurance funds safer. The operational risk of medical insurance is of complexity and uncertainty. And various forms of integration of the basic medical insurance between urban and rural not only increased the difficulty of understanding the risk of medical insurance funds, but also difficult to evaluate and predict the effect of integration. ..Unlike past normal study ways which try to explore the causality, this study is based on the theory of health insurance fund risk to screen factors in operational fund risk and build the model of basic medical insurance risk identification and the warning neural network for rural and urban residents by using Error Back Propagation. In order to debug the model, we use real survey data which are provided by NCMS and medical insurance system of urban and rural residents during the operation as well as various stakeholders in 48 counties in western, middle and eastern areas in China since 2006, so that we can identify the proportion of basic medical insurance fund of urban and rural residents and realize the prediction of the operational fund risk. ..Moreover, this paper uses real data to identify and predict the operational risks in the basic medical insurance fund of urban and rural residents through artificial intelligence techniques, trying to apply multidisciplinary comprehensive evaluation technology from a new perspective to further promote the development of China's basic medical insurance theory, system construction and reform practices by providing theoretical and practical advice.

整合城镇居民基本医疗保险与新农合是医改趋势之一, 它可以改善制度不公平，更重要的是保证医保基金更加安全。医保运行风险具有复杂性和不确定性，多种形式的城乡居民基本医疗保险整合，加重了理解医保基金风险的研究难度，也难以对整合后的效果进行评价和预测。本研究区别于既往探寻因果联系的研究思路，而是基于医保基金风险理论，筛选基金运行风险因素，构建使用反向误差传播算法的城乡居民基本医疗保险风险识别与预警神经网络模型。利用我国东中西部48个县域2006年以来的新农合与城镇居民医保运行及多利益相关方的实地调研数据对模型进行训练，从而实现对医保基金运行风险因素和权重的识别，进而对基金运行风险开展预测。本课题通过人工智能技术并利用现实数据实现对城乡居民基本医疗保险基金运行风险的识别和预测，力求应用新的研究视角和跨学科的综合评价技术，为进一步推进我国基本医疗保险理论发展、制度建设和改革实务提供理论和实证建议。

整合城镇居民基本医疗保险与新农合是医改趋势之一, 它可以改善制度不公平，更重要的是保证医保基金更加安全。医保运行风险具有复杂性和不确定性，多种形式的城乡居民基本医疗保险整合，加重了理解医保基金风险的研究难度，也难以对整合后的效果进行评价和预测。本研究基于医保基金风险理论，筛选基金运行风险因素，构建使用反向误差传播算法的城乡居民基本医疗保险风险识别与预警神经网络模型。通过人工智能技术并利用现实数据实现对城乡居民基本医疗保险基金运行风险的识别和预测，力求应用新的研究视角和跨学科的综合评价技术，为进一步推进我国基本医疗保险理论发展、制度建设和改革实务提供理论和实证建议。本研究显示：经过多年的建设和改革，我国新农合的运行取得了明显成效，新农合覆盖范围不断扩大，保障水平逐渐提高，但部分地区新农合基金存在运行风险。新农合基金风险更可能发生在中西部、人均筹资额低、农民纯收入低、参保人数低、乡镇住院人数占比低、新农合次均住院费用高和县外住院人数占比高的地区。