基于数据特征的电子病历数据驱动临床决策模型研究  

Research on Clinical Decision Model Driven by Electronic Medical Records Based on Data Characteristics

作　　者：杨鑫禹 牟冬梅[1,2] 彭浩 王颖 王萍 李桦[1,2] Yang Xinyu

机构地区：[1]吉林大学公共卫生学院,吉林长春130021 [2]吉林大学第一医院,吉林长春130021

出　　处：《情报理论与实践》2022年第5期181-188,共8页Information Studies:Theory & Application

基　　金：国家自然科学基金项目“信息链视域下电子病历数据驱动健康服务供给侧决策的路径与模式研究”(项目编号:71974074);吉林省科技发展计划项目“面向精准医疗的多模态医学数据融合与挖掘团队”(项目编号:20200301004RQ);吉林大学研究生创新基金资助项目“中文电子病历数据驱动的肺癌分期预测研究”(项目编号:101832020CX274)的成果。

摘　　要：[目的/意义]探索电子病历数据的情报价值,拓展情报学研究数据源和问题域,实现电子病历数据创新使能,为嵌入临床诊疗与临床研究的情报工作和情报服务提供支持。[方法/过程]分析电子病历数据特征及其对数据驱动的影响;针对特征讨论电子病历数据处理技术;基于新技术环境下的信息链理论,构建电子病历数据驱动临床决策模型。实证部分采用随机森林等方法构建电子病历数据驱动的肺癌转移预测模型,实现与验证理论模型。[结果/结论]电子病历数据具有时间依赖、高维稀疏等特征,理论层面基于数据特征构建了电子病历数据驱动临床决策模型,实证层面使用真实电子病历数据构建肺癌转移预测模型,预测模型性能较好。[Purpose/significance]Explore the information value of electronic health records,expand the data source and problem domain of information science research,realize the innovation and enable of electronic health records,and provide support for intelligence work and intelligence services embedded in clinical diagnosis and treatment and clinical research.[Method/process]The paper analyzed the characteristics of electronic health records and its influence on data drive.According to the characteristics,the electronic health records processing technology was discussed.Based on the information chain theory in the new technology environment,a clinical decision model driven by electronic health records was constructed.Finally,the prediction model of lung cancer metastasis driven by electronic health records was constructed by random forest method.[Result/conclusion]Electronic health records were characterized by time dependence and high dimensional sparsity.In theory,a data-driven model of electronic health records oriented to clinical decision making was constructed.In practice,a prediction model of lung cancer metastasis was constructed using real electronic health records,and the prediction model had good performance.

关 键 词：电子病历 数据驱动 决策支持 数据特征 肺癌转移 信息链 

分 类 号：R197.323[医药卫生—卫生事业管理] TP311.13[医药卫生—公共卫生与预防医学]