基于HIS数据的药物相关肌肉不良反应自动监测模块建立与优化

Establishment and optimization of a module for automatic monitoring drug-associated muscle adverse reactions based on HIS

目的:在临床ADE主动监测与智能评估警示系统-Ⅱ中构建药物相关肌肉不良反应(DAMAR)监测模块,为DAMAR的真实世界研究提供智能挖掘工具。方法:依据相关文献和病例资料,建立DAMAR自动监测报警规则,结合机器学习技术构建决策树模型,确定最佳模块设置条件;对住院患者进行监测,验证模块并获得DAMAR发生情况。结果:基于决策树模型的报警规则涉及7个关键词、实验室检验指标和患者手术时间信息等;模块最佳设置条件下阳性预测值由15.31%提升至55.65%,召回率保持100.00%。用于5441例住院患者DAMAR监测,获得阳性患者64例,发生率为1.18%,属于常见药品不良反应(ADR);其中横纹肌溶解严重不良反应患者5例,发生率为0.092%,属于罕见ADR。结论:本研究构建的DAMAR自动监测模块将相关患者手术信息纳入决策树的判断依据,可以更加高效、精准、快速地获取住院人群中的目标病例,为今后开展单药/多药/全药大样本住院患者中DAMAR的真实世界监测与评估研究提供了技术支持。

Objective:To establish an automatic module for drug-associated muscle adverse reaction(DAMAR)in adverse drug event active surveillance and assessment system-Ⅱ,and provide intelligent mining tool for real-world study.Methods:Automatic monitoring alert rules were established based on relevant literatures and medical data and optimal module settings were determined with machine learning techniques to construct a decision tree model.The module was validated by monitoring inpatients and the occurrence of DAMAR was explored.Results:Alarm rules based on the decision tree model included seven keywords,indicators of laboratory test,surgical information,and so on.The positive predictive value and the recall rate of the module were 55.65%and 100.00%respectively at the best optimal module settings.5441 inpatients were monitored for automatic alarms and manual screening of DAMAR and 64 of them were included as positive patients,which was a common adverse drug reaction with the incidence rate of 1.18%,5 patients developed rhabdomyolysis,a rare adverse drug reaction with the incidence rate of 0.092%.Conclusion:The automatic monitoring module for DAMAR established in this study included relevant surgical information of patients into the judgment basis of the decision tree,which could acquire target cases in hospitalized patients more efficiently,accurately,and rapidly,and provide technical support for the real-world study of single-drug/multiple-drug/all-drugs-associated DAMAR in large inpatients.