基于Python的Apriori算法在超长住院患者信息挖掘中的应用  

The Application of Apriori Algorithm based on Python Language in Data Mining of Long Hospitalized Patients

作　　者：刘慧悦[1,2] 杨洋[1,2] 李国垒[3] 金敏[1,2] 张静[1,2] 周柳英[1,2] Liu Huiyue;Yang Yang;Li Guolei;Jin Min;Zhang Jing;Zhou Liuying(Center for Medical Record Management and Information Statistics,Xiangya Hospital of Central South University,Institute of Hospital Management of Central South University,Changsha 410008,Hunan Province,China;不详)

机构地区：[1]中南大学湘雅医院病案管理与信息统计中心 [2]中南大学医院管理研究所,湖南省长沙市410008 [3]中国医学科学院医学信息研究所,北京市100020

出　　处：《中国病案》2022年第6期68-71,共4页Chinese Medical Record

基　　金：中南大学湘雅医院医院管理研究基金项目(2019GL15)。

摘　　要：目的 通过Apriori算法查找某院超长住院日住院病案首页中各指标的关联规则,分析影响超长住院的因素,为降低平均住院日提供思路。方法 采集某院2019年1月1日至2019年12月31日住院天数超过30天的住院病案首页共计3556份,结合2020年版三级医院评审标准中32项医疗安全指标,对住院病案首页中性别、年龄、出院科室、有无手术、医疗安全等21个指标,利用Python语言实现Apriori算法做关联规则分析,分析是否存在关联规则,并分析其原因。结果 共获得强关联规则7203条。普外科出院患者一般进行了手术治疗,其支持度为0.066,置信度为0.741,共235例。总费用15万及以上的住院患者,通常也进行了手术治疗,其支持度为0.234,置信度为0.925,共833例。主要诊断编码在消化系统疾病、恶性肿瘤、循环系统疾病的患者一般有手术治疗,支持度分别为0.055、0.191、0.055,置信度分别为0.875、0.864、0.785,例数分别为196例、680例、556例。结论 通过关联规则分析和发现产生超长住院的原因,关注重点病种及治疗方式、完善绩效考核制度、实行预住院管理模式和运用医联体双向转诊渠道等可以减少患者的住院天数。Objectives To use Apriori algorithm to find the association rules in the front page of medical records of patients with overstay hospitalization in a hospital,in order to analyze the internal reasons of overstay hospitalization,and provide ideas for reducing the average length of stay.Methods A total of 3556 front pages of medical records of patients hospitalized for more than 30 days in a hospital from January 1 st,2019 to December 31 st,2019 were collected.Using Apriori function in package rules of Python language,combined with 32 medical safety indicators in the 2020 version of tertiary hospital evaluation standard,explore whether there were association rules in 21 indicators,including gender,age,discharge diagnosis,surgical treatment,medical safety indicators and so on,of patients with over long hospitalization days in the front page of medical records,and analyze the reasons.Results According to the procedures for the preparation were obtained 7203 strong association rules,the patients discharged from general surgery department generally underwent surgical treatment,with a support of 0.066 and a confidence of 0.741,a total of 235 cases.Inpatients with a total cost of 150000 or more usually underwent surgical treatment,with a support of 0.234 and a confidence of 0.925,a total of 833 cases.Patients with digestive system diseases,malignant tumors and circulatory system diseases generally had surgical treatment.The support degrees were 0.055,0.191 and 0.055 respectively,and the confidence degrees were 0.875,0.864 and 0.785 respectively.The number of cases was 196,680 and 556 respectively.Conclusions Through association rules analysis and finding out the causes of long hospitalization,we could reduce the length of stay of patients by paying attention to key diseases and treatment methods,improve the performance assessment system,implement the pre-hospitalization management model and use the two-way referral channel of medical association.

关 键 词：超长住院日 数据挖掘 PYTHON语言 

分 类 号：R197.323[医药卫生—卫生事业管理]