中药注射剂不良反应随机森林信号模型建立及免疫毒理学评价方法研究

项目名称： 中药注射剂不良反应随机森林信号模型建立及免疫毒理学评价方法研究

项目编号： No.81473514

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： 荆志伟

作者单位： 中国中医科学院

项目金额： 72万元

中文摘要： 中药注射剂风险预警是中药安全性研究的难点，不当使用有悖《内经》有故无殒用药原则，患者反应使预警缺乏可靠方法。目前中药注射剂不良反应(ADR)研究多局限于个案分析和单药信号检测，对复杂用药与个体差异的ADR评价由于统计学方法的局限和缺乏真实世界下的免疫毒理学评估，而导致大量数据混杂偏倚和缺乏定量动态分析，无法实现预警评价。本课题在前期和同期开展重点品种医院集中监测基础上，检测免疫毒理学指标评价患者自身反应，通过追踪访问、专家评定等实现数据规范，运用随机森林模型定量分类判别中药注射剂联合用药（联合药物、给药方式、给药时序）的风险信号，采用条件推断树改进随机森林模型提高预警效能。该方法的建立，有助于识别和预警预测中药注射剂个体化用药的风险，为评价不良事件因果关系和患者反应提供方法学参考，对减少复杂用药而引起的大规模药害事件有重要意义。

中文关键词： 中药注射剂；方法学研究；随机森林；预警方法；免疫毒理学

英文摘要： TCM injections of risk early warning is a key problem for Chinese medicine safety studies, improper use against Neijing is therefore not plunge medicine principle, reaction in patients with early warning and the lack of a reliable method. At present, a lot of traditional Chinese medicine injection adverse drug reactions (ADR) research is limited to signal detection, case analysis and single drug ADR evaluation for complex drug use and individual differences are due to the limitation of statistical methods and lack of immune toxicological assessment under the real world,and lead to big differences among the data and the lack of quantitative dynamic analysis, unable to realize the early warning evaluation. This topic in prophase and at the same time focus on varieties of hospital based on centralized monitoring, detection of toxicology indicator to evaluate the patient's own immune response, by tracking access, expert evaluation, such as data specification, using the random forest model quantitative classification discriminant TCM injections combination(combination of drugs, dosing method and dosing sequence) signal, the risk of using conditions inference tree model to increase the efficiency of the early warning and improve random forest and centralized monitoring data and compared two years ADR report for validation. Early warning and forecast of the establishment of this method helps to identify and the risk of TCM injections individualized medication, to evaluate the causal relationship between adverse events and patient response to provide methodology reference, to reduce complex administration caused by large-scale phytotoxicity events have important significance.

英文关键词： traditional Chinese medicine injection;Methodology;random forest;early warning method;Immune toxicology