过去20年，我国的各个大型医院积累了数以亿计的“电子病历”档案。这些“电子病历”详细记录了病人在医院就诊过程中的大量信息，对它们的深入挖掘具有极高的科学和社会价值。但是，“电子病历”中以自然语言形式出现的中文文本给数据分析带来了极大的挑战。 例如，“对医学文本中的医学术语进行识别和标准化”、“对医学文本中的多维度医学信息进行有效整合以准确重构医学场景”都是医学大数据分析中的核心问题，也是极其困难的问题。本项目力图从统计学角度研究这些问题，建立针对中文电子病历文本分析的统计模型，和能够高效实现“疾病名称”标准化的统计模型，并解决其中的统计推断和统计计算问题。同时，探索构建能够使“医学文本分析”与“医学知识库”高效互动的统计学框架，以更有效地促进医学知识库系统的自我完善和医学自然语言的语意理解。

In past 20 years, millions of Electronic Medical Records (EMRs) have been cumulated in China's hospitals. These EMRs record rich information on what patients have experienced in hospitals, and contain great values. However, the Chinese texts in these EMRs introduce great challenges in data analysis. For example, identification and standardization of medical terminologies and reconstruction of the medical scenario encoded by medical texts are both very challenging but also very important tasks in EMR analysis. In this project, we plan to study these challenging problems from the statistical perspective. We hope to build statistical models for mining Chinese medical texts and standardizing disease names, and investigate the statistical inference and computation of these models. We also want to connect the problem of medical text analysis with medical knowledge database, and establish a intelligent statistical framework with which the two sides can efficiently interact and help each other.