项目名称: 上市后药品不良反应信号检测中双稳健方法的构建
项目编号: No.81502895
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 医药、卫生
项目作者: 张新佶
作者单位: 中国人民解放军第二军医大学
项目金额: 18万元
中文摘要: 充分利用海量自发呈报系统数据,高效、准确地筛选出药品不良反应信号,最大限度地降低“药害事件”对人类的伤害,是目前极受关注并迫切需要解决的问题。课题组在前期进行的一系列药品不良反应信号检测研究中发现,随着大数据技术的发展,自发呈报系统呈现出数据海量性、药品种类多样性以及药品间关系复杂性等新特点,导致现有检测方法出现运算效率低、伪信号多以及定位精确性差等问题。本研究拟在前期工作基础上,利用双稳健思想,探讨一种全新的不良反应信号检测方法。将数据挖掘技术中的boosting算法与贝叶斯logistic回归相结合,最大化提高海量数据的分析效率,并通过控制其他药品与不良反应之间的关系、药品之间的交互作用及其他混杂因素的影响,实现信号的精确定位,有效增强不良反应信号检测的准确性和可靠性。本研究将为上市后药品不良反应监测提供新的方法,促进公共卫生大数据的利用,提高药品风险管理水平、保障居民用药安全。
中文关键词: 药物警戒;双稳健方法;自发呈报系统;信号检测
英文摘要: The importance of detecting postmarketing safety signals earlier and with a high degree of fidelity is increasingly important and of great interest to industry, regulators, and the public. With the development of “big data” technology, spontaneous reporting system presents some new features including huge amounts of data, various drug species and complexity of relationship between drugs. These new features bring big challenge to the current signal detection algorithms, such as low analytic efficiency, confounding and masking effects. To solve these issues, we will propose a new double robust approach by combining boosting algorithm and Bayesian logistic regression. The double robust approach can deal with a large number of drugs simultaneously and address multi-drug interactions and potential confounders. Thus, this new approach may have better signal detection performance to analyze the massive amount of data in the spontaneous reporting system database. Our research will provide a new approach for pharmacovigilance and benefit drug safety surveillance.
英文关键词: pharmacovigilance; double robust approach;spontaneous reporting system;signal detection