引用本文
  • 刘文宝,任东彦,陶峰,陈国良*.基于决策树的住院烧伤患者医疗救治流程优化及规则挖掘[J].第二军医大学学报,2018,39(12):1390-1394    [点击复制]
  • LIU Wen-bao,REN Dong-yan,TAO Feng,CHEN Guo-liang*.Process optimization and rule mining of medical treatment for burn inpatients based on decision tree[J].Acad J Sec Mil Med Univ,2018,39(12):1390-1394   [点击复制]
【打印本页】 【下载PDF全文】 【HTML】 查看/发表评论下载PDF阅读器关闭

←前一篇|后一篇→

过刊浏览    高级检索

本文已被:浏览 370次   下载 227 本文二维码信息
码上扫一扫!
基于决策树的住院烧伤患者医疗救治流程优化及规则挖掘
刘文宝,任东彦,陶峰,陈国良*
0
(海军军医大学(第二军医大学)海军医学系海军卫勤与装备教研室, 上海 200433
*通信作者)
摘要:
目的 探讨基于决策树的归纳分类算法在医疗救治流程优化中的应用。方法 以住院烧伤患者检测结果为基本资料,以医疗救治效率为决策目标,将基于决策树的归纳分类算法运用于医疗救治流程优化,构建其决策树模型,并挖掘出医疗救治流程优化的有用规则。结果 经过决策树流程优化,在10个病理属性中,有4个属性对确定患者的救治方案起到关键作用,即烧伤程度、血生物化学、血压、脉搏。当患者烧伤程度为轻度时,仅需通过考察血生物化学属性即可确定救治方案;当患者烧伤程度为中度时,首先通过考察血生物化学属性,进而再通过考察血压或脉搏属性即可确定救治方案;当患者烧伤程度为重度时直接采用紧急救治方案。结论 以决策树为代表的数据挖掘技术能够较好地辅助烧伤鉴别诊断,优化救治流程。
关键词:  烧伤  临床方案  流程优化  决策树  算法
DOI:10.16781/j.0258-879x.2018.12.1390
投稿时间:2018-04-16修订日期:2018-06-12
基金项目:军队后勤科研重点项目(BWS13C008,BWS17J020).
Process optimization and rule mining of medical treatment for burn inpatients based on decision tree
LIU Wen-bao,REN Dong-yan,TAO Feng,CHEN Guo-liang*
(Department of Naval Health Service and Medical Equipment, Faculty of Naval Medicine, Navy Medical University(Second Military Medical University), Shanghai 200433, China
*Corresponding author)
Abstract:
Objective To explore the application of inductive classification algorithm based on decision tree in optimization of medical treatment process. Methods Taking the test results of the burn inpatients as general data, we used inductive classification algorithm based on decision tree for medical treatment process optimization with medical treatment efficiency as the target. The model of decision tree was constructed and the rules for the optimization of medical treatment process were excavated. Results Among 10 pathological attributes, extent of burn, blood biochemistry, blood pressure and pulse played key roles in determining the patient treatment program after optimizing decision tree process. When the burn was mild, the treatment plan could be determined only by examining blood biochemistry indexes. When the burn was moderate, the treatment plan could be determined first by examining blood biochemistry indexes and then by examining blood pressure or pulse. When the burn was severe, emergency treatment plan should be adopted directly. Conclusion Data mining technology represented by decision tree can contributes to differential diagnosis of burn and optimization of the treatment process.
Key words:  burn patients  clinical protocols  process optimization  decision trees  algorithms