Novel data-mining methodologies for detecting drug-drug interactions : A review of pharmacovigilance literature
Novel data-mining methodologies for detecting drug-drug interactions : A review of pharmacovigilance literature
复制标题
DOI：
--
复制
发表时间：
--
期刊：
影响因子：
--
通讯作者：
I. Heba;A. Amany;S. E. Ahmed;S. Amr
中科院分区：
文献类型：
--
作者：
I. Heba;A. Amany;S. E. Ahmed;S. Amr
对人群健康有重大影响的活动和进行临床研究。PHV的首要目标是及时检测在临床性质、严重程度和/或频率上新颖的不良药物事件(ADE)。直到最近，PHV的核心是通过自发报告系统(SRSS)系统地收集有效的安全数据，这些系统可以作为患者护理的一部分进行严格的分析、解释和操作。数据挖掘算法已经被开发用于从这样的数据库中定量检测ADE的信号。药物相互作用(DDiS)是新药开发和上市后PHV的一个重要问题，占所有ADE的6-30%。因此，本文综述了针对DDIS信号的新的挖掘方法和/或非传统数据源被提出的研究。作者提供了最近的方法创新和用于支持DDIS检测的替代数据来源在上市后期间的重点审查。我们的目的并不是仔细审查所有相关工作。相反，我们简要介绍了基本概念，然后是涉及的数据挖掘算法(DMAS)，涵盖了它们的统计模型、贡献和与DDIS相关的已发表文献的主要发现的计算。在数据挖掘方法论方面，按照数据源轴进行组织。最后，作者提出了一些与当前使用的挖掘算法相关的挑战，并对药物相互作用(DIS)监测的进一步研究提出了建议。
activity with strong implications for population health and conducting clinical research. The overarching goal of PhV is the timely detection of adverse drug events (ADEs) that are novel in their clinical nature, severity, and/or frequency. Until recently, the core of PhV is based on the systematic collection of valid safety data through spontaneous reporting systems (SRSs) that can be rigorously analyzed, interpreted, and acted upon as part of patient care. Data mining algorithms have been developed for the quantitative signal detection of ADEs from such databases. Drug-drug interactions (DDIs) constitute an important problem in the development of new drugs and postmarketing PhV which contribute to 6-30% of all ADEs. This article, therefore, reviews studies in which novel mining approaches and/or nontraditional data sources have been proposed for signaling DDIs. The authors provide a focused review of recent methodological innovations and alternative data sources used to support DDIs detection in the postmarketing period. We do not aim to elaborately examine all relevant work. Instead, we presented a synopsis of basic concepts, then following by the involved data-mining algorithms (DMAs) covering the computation of their statistical models, contributions, and major findings from published literature with respect to DDIs. Regarding data mining methodologies, the review is organized according to data source axis. Finally, the authors presented some of the challenges related to the currently used mining algorithms and suggestions for further research for drug interactions (DIs) surveillance are offered.