基于人工神经网络的冠心病治疗药物验证分析  

Validation and Analysis of Treatment Drugs for Coronary Heart Disease Based on Artificial Neural Network

作　　者：杨帆[1] 许少杰[1] 邹兆重[1] 陈昂[1] 徐泽辉 YANG Fan;XU Shao-jie;ZOU Zhao-zhong;CHEN Ang;XU Ze-hui(Boai Hospital of Zhongshan,Zhongshan 528403,China)

机构地区：[1]中山市博爱医院,广东中山528403

出　　处：《海峡药学》2022年第1期141-144,共4页Strait Pharmaceutical Journal

基　　金：中山市社会公益科技研究项目资金(NO:2018B1100)。

摘　　要：目的利用人工神经网络分析方法,训练、验证临床治疗冠心病用药规律。方法回顾性分析某院2017年至2020年住院治疗的378例冠心病患者的临床数据资料,采用Jupyter Notebook(Python 3.63)构建人工神经网络。结果选取3个异常指标组,328例作为训练集构建模型,50例作为模型验证测试集,验证结果显示,3组平均绝对预测误差分别为8.4%,4.1%以及14.6%,采用敏感度分析挖掘出核心药物。结论人工神经网络验证分析方法可以从临床数据中挖掘出有临床意义的用药规律。OBJECTIVE To use artificial neural network analysis methods to train and verify the clinical medication regulated pattern for coronary heart disease.METHODS Retrospectively analyze the clinical data of 378 patients with coronary heart disease who were hospitalized in a hospital from 2017 to 2020.Jupyter Notebook(Python 3.63)was used to construct an artificial neural network.RESULTS Three abnormal index groups were selected,and 328 cases were used to construct and train the model,and 50 cases were set to verify the model.The verification results showed that the average absolute prediction errors of the three groups were 8.4%,4.1% and 14.6%,respectively,and the core drugs were mined by sensitivity analysis.CONCLUSION The artificial neural network verification and analysis method can dig out clinically meaningful medication regulated pattern from clinical data.

关 键 词：人工神经网络分析 冠心病 用药规律 

分 类 号：R969.4[医药卫生—药理学]