基于比例失衡法联合机器学习算法对艾司氯胺酮不良事件的信号挖掘与分析

扫码查看

原文链接

万方数据
维普

中文摘要：目的挖掘并分析艾司氯胺酮的不良事件信号，为临床安全用药提供参考。方法收集美国食品药品管理局(FDA)不良事件报告系统(FAERS)数据库中2019年第1季度-2023年第4季度的艾司氯胺酮不良事件报告数据，分别采用传统比例失衡法(报告比值比法、信息成分法)和机器学习算法[随机森林(RF)算法、K-近邻算法、极限梯度提升(XGBoost)算法]挖掘艾司氯胺酮不良事件信号，并通过曲线下面积(AUC)评估机器学习算法信号检测结果的准确性。结果共获得5 247条以艾司氯胺酮为首要怀疑药物的不良事件记录，采用传统比例失衡法共检测出138个阳性信号，其中抗胆碱能综合征、尿失禁、复视、肾盂肾炎、自发性气胸、胆道梗阻6个新的不良事件信号未被FDA药品说明书收录，并发现该药可能更容易引发心血管方面的问题。机器学习算法结果显示，XGBoost算法和RF算法性能相对较佳，AUC均值分别为0。928、0。921;共检测出复视、一般身体健康状况恶化、自杀意念、戒断综合征4种新的潜在不良事件信号。结论艾司氯胺酮在获得显著疗效的同时也伴随一些未知风险，临床中可能出现说明书中未提及的不良事件，医疗人员在应用其进行临床治疗时应对相关不良事件保持充分警惕，并及时采取措施保障治疗安全。

外文标题：Signal mining and analysis of adverse events of esketamine based on proportional imbalance method and machine learning algorithms

外文摘要：Objective To explore and analyse the signals of adverse events of esketamine,and to provide references for rational clinical use of the drug.Methods The adverse event reports of esketamine from the first quarter of 2019 to the fourth quarter of 2023 in the U.S.Food and Drug Administration Adverse Event Reporting System(FAERS)database were collected.The reporting odds ratio(ROR)method and information component(IC)method in the disproportionality analysis and random forest(RF)algorithm,K-nearest neighbor algorithm and extreme gradient boosting(XGBoost)algorithm in machine learning algorithms were used for signal mining of target medicines respectively.The accuracy of machine learning signal detection results was assessed by the area under the curve(AUC).Results A total of 5 247 adverse event records with esketamine as the primary suspect drug were obtained.Using the traditional detection measures of dis-proportionality,138 positive signal results were detected,6 new adverse events including anticholinergic syndrome,urinary incontinence,double vision,pyelonephritis,spontaneous pneumothorax,biliary obstruction,were not included in the FDA drug inserts,and it was found that the drug may be more likely to cause cardiovascular problems.The results of the machine learning training showed that XGBoost algorithm and RF algorithm performed moderately well,with AUC means of 0.928 and 0.921,respectively.A total of 4 new potential adverse drug event signals,diplopia,deterioration of general physical health,suicidal ideation and withdrawal syndrome were detected by XGBoost algorithm and RF algorithm.Conclusion Esketamine is accompanied by some unknown risks while obtaining significant efficacy and adverse events not mentioned in the specification may occur in clinical practice.Healthcare professionals should be fully alert to the relevant adverse events when applying them in clinical treatment and take timely measures to ensure the safety of the treatment.

外文关键词：

EsketamineTreatment-resistant depressionAdverse drug eventSignal detectionDisproportional assayMachine learning algorithmFAERS databasePharmacovigilance

作者：

陈曦、刘畅、凌一、张鹤巍、郭晓晶

展开 >

作者单位：

海军军医大学基础医学院(上海 200433)

海军军医大学卫生勤务学系军队卫生统计学教研室(上海 200433)

关键词：

艾司氯胺酮难治性抑郁症药品不良事件信号检测比例失衡法机器学习 FAERS数据库药物警戒

基金：

国家自然科学基金面上项目国家自然科学基金青年科学基金项目上海市公共卫生体系建设三年行动计划优青计划

项目编号：

8207367181703296GWV-10.2-YQ33

出版年：

2024

DOI：

10.12173/j.issn.1005-0698.202408074

药物流行病学杂志

中国药学会武汉医药（集团）股份有限公司

药物流行病学杂志

CSTPCD

影响因子：0.746

ISSN：1005-0698

年,卷(期)：2024.33(9)