项目名称: 诊断试验中面板数据一致性评价的贝叶斯方法研究
项目编号: No.81473069
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 医药、卫生
项目作者: 李婵娟
作者单位: 中国人民解放军第四军医大学
项目金额: 75万元
中文摘要: 临床监测新型设备的研制和更新为临床实践提供了科学价值更高的数据,监测数据构成面板数据,如何根据面板数据评价检测设备的一致性问题已引起各国医药管理机构的重视。目前对于面板数据间的一致性评价方法的研究尚属空白。本项目将以贝叶斯理论为研究基础,研究动态和静态面板数据中一致性评价问题。在构建面板数据一致性评价统计模型的基础上,结合贝叶斯先验分布的构造、后验分布的模拟抽样和估计等问题以及根据不同类型试验终点探讨模型中参数的确定方法,提出动静态面板数据一致性评价统计学指标。并进一步对面板数据一致性评价中样本量的估计、缺失数据的影响及其解决方法和不同横截面与时间维度结构面板数据的一致性评价方法理论等进行研究。最后在理论研究的支持下,通过临床实例数据进行验证和改进,以利于方法体系的推广和应用。
中文关键词: 诊断试验;面板数据;一致性评价;贝叶斯方法
英文摘要: Development and update of new monitoring equipment provides more scientific data in clinical practice. Monitoring data aggregates into a panel data. How to evaluate the consistency of test equipment based on panel data has been attracted attention of national medicine regulatory agencies. However, there still lacks of sufficient theoretical foundations and concrete technical guidelines in agreement assessment about panel data. The project will focus on agreement assessment about static and dynamic panel data based on bayesian methods. On the basis of model construction, we will perform parameter determination considering bayesian prior elicitation, posterior computation and estimation according to different types of endpoints, and explore feasible statistical indices. Furthermore, the project will explore sample size estimation and agreement assessment methods for missing data or panel data with different magnitude of the cross section and time series. In addition, in support of theoretical research, this project aims to promote clinical application based on clinical examples.
英文关键词: diagnosis trial;panel data;agreement assessment;bayesian method