不合理用药危害巨大，及时发现不合理处方是进行有效干预的基础。但目前采用的抽样处方点评模式覆盖面小、漏查率高、监管滞后，亟需改进。本研究拟利用全国合理用药监测系统2010年至今收集的上百亿条门诊处方数据，基于异常检测技术、频繁模式挖掘技术进行多模式多视图的不合理处方识别模型的构建；基于宁波鄞州区数年来积累的临床大数据从多类结局视图角度进行模型可用性的验证并不断迭代优化模型。拟解决三个关键科学问题：多视图不合理处方检测模型构建；多模式方法中不同模型的参数设置；综合判定结局指标的方法。本课题所研究的方法能从大量的处方数据中快速有效地筛选不合理处方，减少抽样点评人力和时间的消耗，具有很好的经济效益；通过发现大量潜在的不合理用药模式，将模型嵌入到日常诊疗流程中，还可实现不合理处方的及时干预，减少不合理用药危害，具有良好的社会效益。这种数据驱动的不合理用药研究也为今后的同类研究提供思路和方法学参考。

To minimize the harm of irrational use of drugs, real-time irrational prescriptions detection is foundation for effective intervention in routine medical practice. Current practice of prescription review need to be improved urgently because of its limitations such as low coverage, high omission rate and lag of supervision. Electronic prescription big data provide a golden opportunity to automate the process of irrational prescription detection. Using billions of prescription data in the Chinese monitoring system for rational use of drugs, the research aims to develop a multi-view and multi-modal irrational prescription detection model based on anomaly detection and frequent pattern mining techniques, and to iteratively evaluate and optimize the model based on multiple outcomes of the clinical data from Yinzhou. The research focuses on three key issues: 1) the construction of a multi-view irrational prescription detection model. 2) the parameter settings of different models in a multi-modal method. 3) the decision making strategy of the model based on the outcomes of the patients. The model aims to effectively filter irrational prescriptions from prescription big data, which is much more economically efficient than manual prescription review. By discovering the latent irrational prescription pattern and imbedding the model into routine medical practice, it is possible to achieve real-time intervention of irrational prescription, which can help to minimize the harm of irrational drug use and provides significant social benefit. Last but not the least, the research can be of great value in providing references for other data-driven studies in the field of irrational drug use.