CAUSAL ANALYSIS OF AES--PATTERN RECOGNITION SOFTWARE

AES--模式识别软件的因果分析

• 批准号: 2637045
• 负责人: DOUGLAS R MACKINTOSH
• 金额: $ 9.99万
• 依托单位: GOOD CLINICAL PRACTICES AUDITING, INC.
• 依托单位国家: 美国
• 财政年份: 1997
• 资助国家: 美国
• 起止时间: 1997-09-30 至 1998-12-15
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/2637045
• 关键词: clinical trials; computer program /software; human data; human therapy evaluation; method development; multiple sclerosis

DESCRIPTION (Adapted from applicant's abstract): Determining if an adverse event is an adverse drug reaction is a difficult process made more complicated in clinical trials that address complex chronic diseases such as immune system compromise. Patients on study typically experience many adverse events, some serious, over the course of drug treatment and after treatment ceases. Previous attempts at facilitating safety analysis have concentrated on simple scoring mechanisms, probability algorithms including a Bayesian approach, recently automated, and global clinical impressions. These analytic methods work best on simple infections and skin diseases, but cannot address complex chronic diseases. In Phase I, software will be developed to analyze clinical trials safety data for chronic diseases. A prototype will be built for multiple sclerosis using actual Phase III trial data. The software will 1) facilitate linkages or pattern recognition among events, between events and medication administration, and between events and lab test results 2) cluster events temporally and by body system and 3) facilitate comparisons between placebo and study drug safety profiles. At the end of Phase I, a panel of clinicians, including medical monitors who prepare safety profiles for clinical trials results, will evaluate the software for ease of use, robustness, and regulatory compliance. In Phase II, the software will be refined and a comparative trial of its analytic power will be conducted on a different chronic disease data set. The goal of the project is to produce software that can be used by clinical researchers, drug and device manufacturers, the FDA, and regulatory agencies in the European Union in determining the relative safety of a new drug.

描述（改编自申请人的摘要）：确定是否 不良事件是药品不良反应是一个艰难的过程， 在临床试验中更复杂， 免疫系统受损等疾病。 参加研究的患者通常 在药物治疗过程中， 治疗结束后和治疗结束后。 以前为便利 安全性分析集中在简单的评分机制上， 概率算法，包括贝叶斯方法，最近 自动化和全球临床印象。 这些分析方法 对简单的感染和皮肤病效果最好，但不能解决 复杂的慢性病。 在第一阶段，将开发用于分析临床试验安全性的软件 慢性病的数据。 一个原型将建立多个 使用实际的III期试验数据。 该软件将1） 促进事件之间的联系或模式识别 以及事件与实验室检查结果之间的关系 2)按时间和身体系统对事件进行聚类，以及3）促进 安慰剂和研究药物安全性特征之间的比较。 年底 在第一阶段，一个临床医生小组，包括医疗监测员， 准备临床试验结果的安全性特征，将评估 易于使用、健壮性和法规遵从性的软件。 在 在第二阶段，将对软件进行改进，并对其进行比较试验。 将对不同的慢性病数据进行分析 集 该项目的目标是生产可以使用的软件 临床研究人员、药物和器械制造商、FDA，以及 欧盟的监管机构在确定相对 新药的安全性。