|
|
何跃, 王爱欣
, 丰月, 王莉
四川大学商学院 成都 610065
He Yue, Wang Aixin, Feng Yue, Wang Li
中图分类号: TP399
通讯作者:
收稿日期: 2017-09-18
修回日期: 2017-10-17
网络出版日期: 2018-01-25
版权声明: 2018 《数据分析与知识发现》编辑部 《数据分析与知识发现》编辑部
展开
摘要
【目的】随着门诊的日就诊人数逐渐增多, 优化门诊药房药品摆放布局, 能够有效提高整个药房系统的服务效率。【方法】选择处方数量最多的两个科室的处方数据, 应用K-means聚类算法将数据集划分为4个子数据集, 使用Apriori算法对4个子数据集进行关联规则挖掘, 得到31条药品有效规则和18条药类有效规则。【结果】综合药类和药品有效规则中挖掘出的信息, 结合国家药品储存陈列规范, 在得到某医院门诊药房的药房管理专家认可的情况下, 设计出药类和药品的大致布局。【局限】只提取两个科室的处方数据, 用于关联规则分析的处方数据不够完善。【结论】将关联规则方法和K-means聚类算法应用于解决门诊药房的药品陈列布局问题, 用数据支撑药品陈列布局设计, 并得到药房专家的认可。有利于减轻药剂师的工作强度, 缩短患者取药时间, 提高整个药房的服务效率。
关键词：
Abstract
[Objective] As the number of outpatient visits increases, optimizing the layout of pharmacy drugs can improve its service efficiency. [Methods] Firstly, we chose two departments with the largest number of prescriptions, which were divided into four sub groups with the K-means clustering method. Then, we used Apriori algorithm to explore the association rules among them. Finally, we obtained 31 effective drug layout rules and 18 effective drug class rules. [Results] We designed general layout rules for prescription drugs based on the collected data along with national drug storage and display standards, which were approved by the experts. [Limitations] We only studied prescription records from two departments, which might not yield the best association rules. [Conclusions] The proposed method could reduce the workload of pharmacists and the waiting time of patients, which improve the pharmacy services.
Keywords：
门诊是医院中不可缺少的组成部分之一, 作为医院开放的、直接对外提供服务的窗口, 其运营的好坏, 服务质量的高低, 涉及到多个方面, 如环境是否安静整洁, 各通道是否拥堵, 患者排队等待时间是否过长, 工作人员能否高效准确地为患者发药等, 这些都会影响到患者对医院的整体感受, 甚至影响到医院的声誉。而门诊药房是医院门诊部门的一个重要组成部分, 是医院为门诊患者提供医疗服务的最后一个环节, 患者在经历了挂号、缴费、就诊等环节长时间的等待后, 其心理情绪已处于消极状态, 若门诊药房排队取药的等待时间过长, 更容易造成患者的不满。门诊药房作为一个为患者提供药品消费的服务系统, 除了具有一般商品服务系统的特征之外, 还有其特有的特点: 首先, 门诊药房销售的商品(药品)是一种特殊的产品, 它与人们的健康密切相关, 如果调剂出现错误, 会对病人的健康产生威胁, 这就要求药剂师要具备良好的专业素质和崇高的职业精神。其次, 药房服务质量的好坏在很大程度上是由患者的感知决定的, 所以在进行服务时应考虑多个因素, 如等候区设置、服务流程、处理问题的能力、服务态度和服务人员的亲和力等。因此, 提高门诊药房的运行效率以及患者对门诊药房服务的满意度, 是研究的重要方面之一。
以某医院为例(简称AB医院), 由于每天门诊患者多, 人流量大, 造成患者取药等待时间过长, 对药房服务不满; 再加上药房空间狭长, 布局不合理, 很多药品的陈列是按照药理学分类摆放, 未考虑各类药品之间的联系, 使得同一张处方上的药品可能存在距离较远的不同货架上, 造成药房工作人员劳动强度大, 员工抱怨较多。因此, 选择AB医院的门诊药房作为研究对象, 具有重要的学术意义和实践意义。具体表现在: 第一, 通过优化药房布局, 可以缩短药剂师调剂时间, 减轻工作强度, 同时可以为患者取药节约时间, 减少患者对药房服务的不满。第二, 针对门诊药房中存在的复杂问题, 若能得到有效解决, 可为其他医院提供参考价值。
药房在人类社会中必不可少, 它肩负着维持人们身体健康的重要使命, 国内外学者对药房进行了多方面的研究。
有关关联规则和聚类分析在药房管理中的应用, Khader等[1]提出用数据挖掘方法以加强药房的自动化管理, 将关联规则算法之一的FP-growth方法应用于中心药房的处方数据库中, 以研究处方药品之间的关联。Song等[2]采用聚类分析、德尔菲分析和决策分析等方法分析了各种紧急事件的数据和与事件相关的病情细节, 以确定最佳药物组合, 为紧急药物治疗提供了有效的参考价值。贾克斌等[3]提出一套创新型数据采集与挖掘方案, 选用Apriori算法对产科数据集展开关联分析, 分析结果对于药房的备药、发药, 手术室准备, 医护人员排班等具有指导意义和参考价值。林淑芳[4]运用关联规则方法对药品日常销售产生的数据进行数据挖掘和相关分析, 从而分析和判断顾客的购物规律和趋势, 为完善零售经营决策方案提供有价值的信息支持。王文青等[5]建立了基于聚类分析方法的药品快速配药区, 提高了门诊药房的配药效率。
有关药房布局的研究中, Lester等[6]采用链接分析方法了解药房的物理布局对药剂师执行任务的影响。结果表明, 链接分析可以识别药剂师在什么样的物理环境中可以更好地完成任务。McDowell等[7]使用加权评分系统评估医院药房的布局设计, 以确定在可行性、安全性、灵活性、运输距离、可维护性、环境等方面优于其他的药房布局。李秀敏等[8]提出从优化药房空间布局、调整药师工作岗位和时间安排、完善取药和等候环境等方面入手优化门诊药房工作流程, 从而缩短患者取药等待时间, 提高其对门诊药房的满意度。张婷等[9]通过使用门诊药房自动化系统, 调整了药房布局、对上机药品进行调试优化、改进加药和盘点操作、制订了门诊药房合理排班作业内容、建立了完整的药品有效期管理方案。
有关药房服务的研究中, Surur等[10]研究了客户对门诊药房服务的满意度, 结论表明, 门诊药房客户的总体平均满意度较低, 并且在不同的社会人群之间存在差异。Poulin等[11]实施了各种质量改进举措(这些举措被分为环境、人员、交际和技术), 提高了VA医疗中心门诊药房服务的客户满意度。从商业的角度出发, Burger等[12]则认为如果医院建造和经营的门诊药房想要成功, 其服务应该作为竞争性的业务运行。
此外, 很多学者在对布局问题的研究中应用关联规则方法。Chandanan等[13]认为关联规则在营销和零售战略中很有用, 它还有许多潜在的应用领域, 如设计、布局和客户隔离等。Zulfikar等[14]在帮助零售公司促销商品中应用了关联规则算法, 并且已经建立的关联规则还可用作确定商品布局的参考。杨丰梅等[15]认为基于关联规则的购物篮分析对实体或在线零售企业改善品类管理、优化门店布局陈列、实现关联促销与推荐等起着重要作用。
但是, 有关关联规则和聚类分析在门诊药房布局问题中的应用研究并不多, 相关研究中也很少提出解决药房布局问题的具体方法。本文将关联规则方法和K-means聚类算法应用于AB门诊药房布局问题中, 找出药品和药类的有效规则, 根据得到的结果设计有关药类及药品的布局方案, 有利于提高整个药房系统的服务效率, 同时提高药房工作人员和患者的满意度。
AB医院是较早引进信息化管理的单位, 在其数据库中积攒了各部门大量的数据, 如住院部、急诊部和门诊部等部门日常工作中的一些信息。隶属于门诊部门的门诊药房也是医院一个重要的组成单位, 药房中销售的各种药品的详细信息, 及每天处理的处方信息也都记录在数据库中。
AB门诊部科室众多, 每个科室每天服务的患者数量多, 致使门诊药房每天处理的处方量相当庞大, 对数据库中所有科室的所有处方信息加以分析, 难度较大。因此, 首先对药房进行调研, 手工随机采集一些处方信息(称为样本数据), 并对样本数据进行分析, 找出处方数量最多的两个科室, 然后向AB医院申请与两个科室有关的更多数据。
申请到的数据还需进行进一步的清洗处理, 如剔除对聚类分析和关联规则分析都无用的单种药品处方等, 保留研究所需的数据。
聚类分析的研究已有多年历史, 是比较成熟的研究方法, 产生了大量的经典算法, 主要可以分为5大类: 划分方法[16]、层次的方法[17]、基于密度的方法[18]、基于网络的方法[19]和基于模型的方法[20]。
本文拟用聚类方法对庞大的数据集进行划分, 而K-means算法[21]是典型的划分方法。因此, 使用K-means聚类算法对门诊处方中常一起销售的药品进行聚类, 每个药品代表一个数据对象, 药品之间的共同销售次数代表数据对象之间的距离, 将距离大的数据对象归为同一类。这样众多药品就被划分为较小的子类, 便于后续分析。
关联规则是寻找事务项之间的联系规律和结构特征的主要数据挖掘方法。最早用于商场购物篮分析问题[22], 这与药房中药品的销售类似, 如果能将经常联合使用的药品摆放在一起, 就可以节约药剂师的调剂时间, 减少调剂时所走的路程, 减轻工作人员的劳动强度, 提高药房的服务效率。
Apriori算法[23]是关联规则挖掘领域的经典广度优先算法。Apriori算法在关联规则挖掘过程中, 会生成大量候选项集, 再对其进行剪枝。然而, 在此过程中生成的候选项并不是都有效, 有些甚至根本不会出现在事务项目集中。另外, 候选项集的生成过程会带来较大的系统开销, 数据库与内存之间进行不断的信息交流, 降低算法的处理效率。在1995年, Savasere等[24]提出将数据分割应用于Apriori算法的思想, 可有效改善关联规则挖掘在处理大容量数据集时内存不足的问题。其基本思想是: 将原始大容量的事务数据集, 分为若干个小容量、独立的、互不相交的较小事务集; 再对每个小事务集用现有的关联规则算法进行挖掘; 将所有的局部关联规则中涉及到的项合并起来, 再对其进行整体的数据集扫描, 以相同的最小支持度和最小置信度进行全局数据挖掘, 避免遗漏。
本文利用Apriori算法分别对聚类分析所得的子类进行关联规则挖掘。
通过对药房进行调研, 共采集到1 740张处方样本数据, 对样本数据进行分析可得, 风湿免疫科和中西医结合科的处方量较大, 两者的总量占样本处方总量的21.7%。因此, 向AB医院申请上述两个科室开具的处方, 用于后续分析, 时间段为2016年每个月5号、12号、15号、20号、25号, 一共60天。数据中包括有: 处方号、性别、开单科室、药品医保类型、病人类型、药品规格、数量、价格等信息, 共109 669条记录。经统计, 原数据中处方总量为48 837张, 单种药品处方21 511张, 多种药品处方27 326张。而单种药品处方对聚类分析和关联规则分析都无用, 属于无效数据, 应该剔除。由于研究只需要其中的处方号、药品编码、药品类别信息, 所以采用Java语言编写程序对原数据进行清洗, 共得到88 148条有效记录。
在27 326张多种药品处方中, 两种药品联合使用次数在1 000次以上的有7对药品, 有20对药品的联合使用次数在500次以上。因此, 构建了818种药品的相似对角矩阵, 如表1所示, 矩阵中两种药品交叉的表格位置的数值表示这两种药品联合使用的次数。
表1 联合使用药品的相似对角矩阵示例
| 药品编号 | 120209Abm3010 | 140108Aiv1690 | … | 110101Abk1605 | 120502M811682 |
|---|---|---|---|---|---|
| 120209Abm3010 | 0 | 493 | … | 412 | 2490 |
| 140108Aiv1690 | 493 | 0 | … | 66 | 465 |
| …… | …… | …… | … | …… | …… |
| 110101Abk1605 | 412 | 66 | … | 0 | 218 |
| 120502M811682 | 2490 | 465 | … | 218 | 0 |
利用Clementine软件中K-means聚类算法对上述所得的相似对角矩阵中的药品按共同销售次数进行聚类分析后得到4个子类。第1类至第4类分别有210, 203, 203和202种药品。其平均销售次数如表2所示。这4个子类内部药品的联合使用平均次数都在12次以上, 各子类间的药品联合使用的现象较少, 其平均次数都低于0.1次。
表2 4个子类平均共同销售次数及子类间的平均共同销售次数
| 第1类 | 第2类 | 第3类 | 第4类 | |
|---|---|---|---|---|
| 第1类 | 16.252 | 0.047 | 0.085 | 0.041 |
| 第2类 | 0.047 | 12.301 | 0.006 | 0.002 |
| 第3类 | 0.085 | 0.006 | 12.376 | 0.004 |
| 第4类 | 0.041 | 0.002 | 0.004 | 12.338 |
使用Clementine软件中Apriori算法分别对聚类分析所得的4个子类进行关联规则挖掘。第1类至第4类的记录数量分别为26 044, 11 529, 12 350和11 054条。将最小支持度设置为1.0%, 最小置信度设置为50.0%。从第1类中得到23条规则, 如表3所示。
表3 第1类关联规则分析结果一览表
| 规则 ID | 后件 | 前件 | 前件支持度 (%) | 置信度 (%) | 规则支持度 (%) | ||
|---|---|---|---|---|---|---|---|
| 1 | 020207A532305 = 1 | 110101Abk1605=1 | and | 120209Abm3010 = 1 | 1.559 | 94.581 | 1.474 |
| 2 | 020207A532305 = 1 | 110101Abk1605 = 1 | 4.362 | 92.782 | 4.047 | ||
| 3 | 120209Abm3010 = 1 | 140103A180151 = 1 | 1.014 | 73.485 | 0.745 | ||
| 4 | 090123M092310 = 1 | 040266AAU0186 = 1 | and | 150304MCX2705 = 1 | 1.014 | 69.318 | 0.703 |
| 5 | 120209Abm3010 = 1 | 150403Abo1655 = 1 | and | 020207A532305 = 1 | 1.286 | 68.060 | 0.875 |
| 6 | 120209Abm3010 = 1 | 140234A568418 = 1 | and | 120502M811682 = 1 | 1.425 | 66.038 | 0.941 |
| 7 | 090140A4421AV = 1 | 090128M0907CE = 1 | 1.375 | 63.966 | 0.879 | ||
| 8 | 120209Abm3010 = 1 | 140234A182363 = 1 | and | 120502M811682 = 1 | 2.607 | 63.918 | 1.666 |
| 9 | 120209Abm3010 = 1 | 150403Abo1655 = 1 | and | 120502M811682 = 1 | 2.964 | 63.860 | 1.893 |
| 10 | 120502M811682 = 1 | 090140A4421AV = 1 | and | 120209Abm3010 = 1 | 3.390 | 61.608 | 2.089 |
| 11 | 120209Abm3010 = 1 | 150403Abo1655 = 1 | 6.385 | 60.794 | 3.882 | ||
| 12 | 040303Abd16Ae = 1 | 160002A951451 = 1 | 1.317 | 60.058 | 0.791 | ||
| 13 | 2704CEM172120 = 1 | 2704bnK251626 = 1 | 1.421 | 58.760 | 0.837 | ||
| 14 | 160002A951451 = 1 | 040303Abd16Ae = 1 | 1.363 | 58.028 | 0.791 | ||
| 15 | 090123M092310 = 1 | 040266AAU0186 = 1 | and | 270462Afk0612 = 1 | 1.490 | 57.216 | 0.852 |
| 16 | 120209Abm3010 = 1 | 090140A4421AV = 1 | and | 120502M811682 = 1 | 3.717 | 56.198 | 2.089 |
| 17 | 040271Abd2045 = 1 | 270462Afk0612 = 1 | and | 120502M811682 = 1 | 2.020 | 55.133 | 1.114 |
| 18 | 150304MCX2705 = 1 | 040281M252912 = 1 | 2.861 | 54.497 | 1.559 | ||
| 19 | 120209Abm3010 = 1 | 150403Abs1655 = 1 | and | 120502M811682 = 1 | 2.020 | 52.281 | 1.056 |
| 20 | 020207A532305 = 1 | 040271Abd2045 = 1 | and | 120209Abm3010 = 1 | 1.286 | 51.940 | 0.668 |
| 21 | 040266AAU0186 = 1 | 090123M092310 = 1 | and | 270462Afk0612 = 1 | 1.651 | 51.628 | 0.852 |
| 22 | 120209Abm3010 = 1 | 120502M811682 = 1 | 18.572 | 50.941 | 9.461 | ||
| 23 | 120502M811682 = 1 | 150403Abs1655 = 1 | and | 120209Abm3010 = 1 | 2.100 | 50.274 | 1.056 |
由表3可知, 药品有效的关联规则中, 前件的药品数量不超过两种, 其中前件中有两种药品的规则有14条, 前件中只有一种药品的规则有9条。产生这种现象的主要原因是风湿免疫科和中西医结合科这两个科室联合用药的数量通常为2-4种。从规则前件支持度的角度看, 第22条规则的前件支持度最高, 为18.572%, 其次是第11、第2条规则, 它们的前件支持度分别为6.385%、4.362%。据专家解释, 各规则前件支持度普遍偏低是由多方面原因引起的, 如药理作用和主治功能相同的药品可能由不同的厂家生产, 并对不同厂家生产的药品赋予不同的药品编号, 即一种药品的替代药品很多。从规则置信度的角度看, 第1条规则的置信度最高, 为94.581%, 这表明, 如果在一张处方中包含有编号为110101Abk1605的药品和编号为120209Abm3010的药品, 那么该处方中包含有编号为020207A532305的药品的可能性为94.581%, 即这三种药品联合使用的几率很高。其次, 第2条规则的置信度为92.782%, 居第二, 并且该规则的前件置信度也较高, 是一条很有价值的规则。从这条规则的支持度看, 规则支持度最高的是第22条规则, 为9.461%, 其余药品规则的规则支持度都较低, 但这种现象与规则前件支持度是吻合的。
对第2类药品进行关联规则分析, 共挖掘得到7条有效规则, 如表4所示。可知第7条规则的前件支持度最高, 为4.971%, 其次是第3条规则, 前件支持度为3.466%。第1条规则的置信度最高, 为100.000%, 这表明, 如果一张处方中包含编号为070501B800493的药品, 那么该处方中肯定包含编号为030202BIO2547的药品。
表4 第2类关联规则分析结果一览表
| 规则 ID | 后件 | 前件 | 前件支持度 (%) | 置信度 (%) | 规则支持度 (%) |
|---|---|---|---|---|---|
| 1 | 030202BIO2547 = 1 | 070501B800493 = 1 | 1.766 | 100.000 | 1.766 |
| 2 | 070501B800493 = 1 | 030202BIO2547 = 1 | 2.027 | 87.097 | 1.766 |
| 3 | 160011B240270 = 1 | 160011B250270 = 1 | 3.466 | 73.585 | 2.551 |
| 4 | 140105A742341 = 1 | 140106A141815 = 1 | 1.112 | 70.588 | 0.785 |
| 5 | 010109MCE1231 = 1 | 010304A560403 = 1 | 1.766 | 70.370 | 1.243 |
| 6 | 010202A181696 = 1 | 090559A262552 = 1 | 2.158 | 63.636 | 1.373 |
| 7 | 160011B250270 = 1 | 160011B240270 = 1 | 4.971 | 51.316 | 2.551 |
对第3类药品进行分析, 只得到1条有效规则。该规则的前件为编号为271304M640735的药品, 后件为编号为271308U752526的药品, 其前件支持度为1.660%, 规则置信度为64.103%, 规则支持度为1.064%。第4类药品关联分析没有挖掘出置信度高于50%的规则。
药品有效规则31条, 涉及到的药品共有33种。为了防止子类之间的药品有效规则的丢失, 将这些药品结合起来做关联规则分析。选择与这些药品相关的记录, 并在Clementine中做相同的工作, 此模型没有得到任何有效规则, 这说明将海量数据的原始数据集划分成子类数据的过程中, 没有丢失原始数据集中的有效规则。
AB门诊药房中药品种类繁多, 一种功能相同的药品, 可能在规格、剂型、生产厂家等方面都不相同, 这些因素也是影响药房药品陈列的重要因素。因此, 找出有效的药类之间的关联规则也很重要。风湿免疫科和中西医结合科这两个科室常用的药共有23类。药类关联规则挖掘, 发现了18条药类之间的规则, 如表5所示, 其中, 第7条规则的前件支持度和规则支持度最高, 分别为25.366%和14.896%; 第1条规则的置信度最高, 为75.804%。
表5 药类关联规则分析结果一览
| 规则 ID | 后件 | 前件 | 前件支持度 (%) | 置信度 (%) | 规则支持度 (%) |
|---|---|---|---|---|---|
| 1 | 抗肿瘤药 = 1 | 血液系统药物 = 1 and 镇痛、解热、抗炎、抗痛风药 = 1 | 2.164 | 75.804 | 1.640 |
| 2 | 镇痛、解热、抗炎、抗痛风药 = 1 | 调节水盐、电解质及酸碱平衡药 = 1 | 1.710 | 73.233 | 1.252 |
| 3 | 维生素矿物质类及肠内肠外营养药 = 1 | 激素类及影响内分泌药物 = 1 and 消化系统药物 = 1 | 6.963 | 62.986 | 4.386 |
| 4 | 抗肿瘤药 = 1 | 血液系统药物 = 1 and 消化系统药物 = 1 | 1.592 | 62.069 | 0.988 |
| 5 | 抗肿瘤药 = 1 | 血液系统药物 = 1 and 激素类及影响内分泌药物 = 1 | 2.892 | 59.367 | 1.717 |
| 6 | 激素类及影响内分泌药物 = 1 | 心血管系统药物 = 1 and 维生素矿物质类及肠内肠外营养药 = 1 | 1.735 | 59.072 | 1.025 |
| 7 | 维生素矿物质类及肠内肠外营养药 = 1 | 激素类及影响内分泌药物 = 1 | 25.366 | 58.724 | 14.896 |
| 8 | 抗肿瘤药 = 1 | 血液系统药物 = 1 | 7.358 | 57.761 | 4.250 |
| 9 | 抗肿瘤药 = 1 | 血液系统药物 = 1 and 免疫系统药物 = 1 | 1.651 | 56.984 | 0.941 |
| 10 | 中成药 = 1 | 抗感染类 = 1 and 呼吸系统药物 = 1 | 1.102 | 56.811 | 0.626 |
| 11 | 维生素矿物质类及肠内肠外营养药 = 1 | 激素类及影响内分泌药物 = 1 and 消化系统药物 = 1 and 免疫系统药物 = 1 | 1.940 | 55.472 | 1.0763 |
| 12 | 中成药 = 1 | 治疗精神障碍药物 = 1 | 1.267 | 53.468 | 0.677 |
| 13 | 激素类及影响内分泌药物 = 1 | 血液系统药物 = 1 and 维生素矿物质类及肠内肠外营养药 = 1 | 2.834 | 53.360 | 1.512 |
| 14 | 维生素矿物质类及肠内肠外营养药 = 1 | 血液系统药物 = 1 and 激素类及影响内分泌药物 = 1 | 2.892 | 52.278 | 1.512 |
| 15 | 抗肿瘤药 = 1 | 血液系统药物 = 1 and 维生素矿物质类及肠内肠外营养药 = 1 | 2.834 | 51.809 | 1.468 |
| 16 | 维生素矿物质类及肠内肠外营养药 = 1 | 抗肿瘤药 = 1 and 激素类及影响内分泌药物 = 1 | 6.644 | 51.185 | 3.401 |
| 17 | 中成药 = 1 | 调节水盐、电解质及酸碱平衡药 = 1 and 镇痛、解热、抗炎、抗痛风药 = 1 | 1.252 | 51.170 | 0.641 |
| 18 | 激素类及影响内分泌药物 = 1 | 血液系统药物 = 1 and 抗肿瘤药 = 1 and 维生素矿物质类及肠内肠外营养药 = 1 | 1.468 | 50.873 | 0.747 |
同AB门诊药房的药房管理专家对上述所得到的药类有效规则和药品有效规则进行讨论, 这些有效规则的价值得到了他们的肯定。
AB门诊药房按照国家规定的药品陈列规范, 分类对药品进行陈列。在不违反药品陈列规范的前提下, 先从药类的陈列入手, 将关联性强的药类相邻陈列; 再从药品陈列的角度看, 将同类药品中关联性强的药品相邻摆放; 将相邻药类之间关联性强的药品尽量近距离摆放。
(1) 药类陈列分析
如表5所示, 在药类有效规则中, 涉及到13类药物, 其中, 激素类及影响内分泌药物、维生素矿物质类及肠内肠外营养药、血液系统药物、抗肿瘤药这4类药物在这些规则中出现的次数最高, 分别是9次、9次、9次、8次。这4类药物之间也有紧密的联系。
在第7条药类有效规则中, 如果在处方中包含有激素类及影响内分泌药物, 那么该处方中同时包含有维生素矿物质类及肠内肠外营养药的概率为58.724%。这两类药物还同时作为其他7条规则前后件的一部分出现过, 它们分别是第3、6、11、13、14、16、18条药类有效规则。因此, 激素类及影响内分泌药物与矿物质类及肠内肠外营养药应相邻陈列。此外, 由第8条药类有效规则可知, 血液系统药物与抗肿瘤药应相邻陈列。
激素类及影响内分泌药物、维生素矿物质类及肠内肠外营养药、血液系统药物、抗肿瘤药在第5、13、14、15、16条药类有效规则的前后件中, 都仅包含有这4类药物的其中3类, 在第18条规则中仅涉及到这4类药物。因此, 这4类药物应就近陈列。
在第1、4、6、9、11条药类有效规则的前件中, 免疫系统药物、消化系统药物、心血管系统药物、镇痛、解热、抗炎、抗痛风药同上述4类高频药物中的血液系统药物、激素类及影响内分泌药物、维生素矿物质类及肠内肠外营养药同时出现, 因此, 可以将这4类药物陈列在4类高频药物的附近。
通过分析规则2和规则17可得调节水盐、电解质及酸碱平衡药与镇痛、解热、抗炎、抗痛风药应相邻陈列, 并将它们作为一个整体靠近中成药摆放。由规则10和规则12可知抗感染类、呼吸系统药物、治疗精神障碍药物都应向中成药靠近陈列。
(2) 药品摆放分析
为了分析有效的药品规则, 首先在AB门诊药房中查询这33种药品的名称, 结果如表6所示。根据表3、表4和表6, 发现叶酸片与甲氨蝶呤片应并排摆放在血液系统药物与抗肿瘤药物的分界线边缘; 醋酸泼尼松片、阿法骨化醇软胶囊以及碳酸钙D3片应该在激素类及影响内分泌药物与维生素矿物质类及肠内肠外营养药的边界处就近摆放; 此外, 碳酸氢钠片和苯溴马隆片应近靠调节水盐、电解质及酸碱平衡药与镇痛、解热、抗炎、抗痛风药的边界线并排摆放。最后, 同一类药物内部各药品应并排摆放在一起。
表6 有效规则中药品所对应的药类一览表
| 药品编号 | 通用名 | 药品类别 |
|---|---|---|
| 010109MCE1231 | 阿莫西林胶囊 | 抗感染类 |
| 010202A181696 | 拉米夫定片 | 抗感染类 |
| 010304A560403 | 甲硝唑片 | 抗感染类 |
| 020207A532305 | 甲氨蝶呤片 | 抗肿瘤药 |
| 030202BIO2547 | 盐酸利多卡因注射液 | 麻醉药及麻醉辅助用药 |
| 040266AAU0186 | 氯诺昔康片 | 镇痛、解热、抗炎、抗痛风药 |
| …… | …… | …… |
| 160011B250270 | 氯化钠注射液(软袋)(250ml) | 调节水盐、电解质及酸碱平衡药 |
| 270462Afk0612 | 荧光素钠注射液 | 中成药 |
| 2704bnK251626 | 蓝芩口服液 | 中成药 |
| 2704CEM172120 | 感咳双清胶囊 | 中成药 |
| 271304M640735 | 生脉胶囊 | 中成药 |
| 271308U752526 | 复方丹参滴丸 | 中成药 |
(3) 药类与药品结合陈列设计
综合上述药类和药品有效规则中挖掘出的信息, 结合国家药品储存陈列规范, 药品需分类摆放, 易混淆药品不能摆放在一起等, 在得到AB门诊药房的药房管理专家认可的情况下, 本文设计出有效规则中所设计的药类和药品的大致布局, 如图1所示。
本文将关联规则方法和K-means聚类算法应用于解决门诊药房的药品陈列布局问题中, 挖掘出31条药品有效规则和18条药类有效规则, 根据这些规则设计了药房药品的优化布局, 将销售量最多的药类陈列于中间位置, 将经常一起销售的药类相邻陈列, 相邻药类或同一药类中经常一起销售的药品就近摆放。有利于降低药剂师调剂所花费的时间, 减轻他们的工作强度, 缩短患者取药时间, 提高整个药房的服务效率。本文的创新之处在于:
(1) 由于涉及药品信息的数据集太大, 在进行关联规则分析之前, 利用聚类方法将数据集划分成较小的子数据集, 再对每个子数据集进行关联规则数据挖掘, 最后将子数据集中得到的有效规则中涉及的药品再做一次关联规则分析, 防止药品有效规则的遗漏。
(2) 目前将关联规则数据挖掘方法应用于医院门诊药房的案例较少, 而从药类层面进行关联规则分析, 优化药房布局的研究思路几乎没有。本文利用数据支撑了药品陈列布局, 并得到药房管理专家的认可。
由于AB门诊药房每天的处方量太大, 该研究没有提取所有科室的处方数据, 只提取了AB医院的风湿免疫科和中西医结合科两个科室的门诊处方数据, 因此, 用于关联规则分析的处方数据不够完善。虽然分析结果得到药房管理专家的认可, 但还不够全面。在未来, 将研究更多科室的处方数据以解决门诊药房的布局问题, 还可从科室的角度研究药品的销售情况, 看是否有必要为某些科室所涉及到的药品设立专门的分区, 以及科室之间交叉用药的程度等。
何跃: 提出研究方向, 论文修订;
王爱欣: 提出研究思路, 设计研究方案, 撰写论文;
丰月: 采集、清洗和分析数据;
王莉: 论文最终版本修订。
所有作者声明不存在利益冲突关系。
支撑数据由作者自存储, E-mail: wax199400@foxmail.com。
[1] 王爱欣. data.zip. 风湿免疫科、中西医结合科药品数据明细.
| [1] |
Pharmacy Robotic Dispensing and Planogram Analysis Using Association Rule Mining with Prescription Data [J].https://doi.org/10.1016/j.eswa.2016.02.045 URL [本文引用: 1] 摘要
Automation in pharmacies has achieved innovative levels of effectiveness and savings. In the present day, automated pharmacies are facing extremely large demands of prescription orders specifically at the central fill pharmacies that distribute drugs to retail pharmacies. As a result, improvements are necessary to the Robotic Prescription Dispensing System (RPDS) and RPDS planogram to increase the throughput of prescriptions. RPDS planogram defines where to allocate the dispensers inside the robotic unit and how to distribute them among the multiple robotic units. This research utilizes the pharmacy prescriptions database to extract useful knowledge to improve different strategies in pharmacy automation by using a data mining approach. In this study, a data mining tool is proposed to enhance pharmacy automation. Frequent Pattern Growth (FP-growth) approach, which is one of the algorithms of Association Rule Mining (ARM), is applied to an actual prescriptions database of a central fill pharmacy to study the associations within the prescribed drug regime. The FP-growth application in a prescriptions database is novel; thus, FP-growth is tested on both sequential mode, and parallel mode by using a distributed platform Hadoop and MapReduce paradigm. Two types of association rules are extracted: 1) associations among different drugs that involve their different dosage strengths and manufacturers; and 2) associations that include only information about different drug generic and brand names. The importance of the extracted association rules is evaluated by the use of different measures, including the support, confidence, lift and conviction. The discovered rules disclose strong associations among the purchased drugs that improve the allocation and distribution of dispensers among the robotic units, in addition to enhancements in other pharmacy managerial strategies.
|
| [2] |
Practical ‘Modular Design’ Research of Emergency Drug Supplies in Hospitals [J].https://doi.org/10.1136/ejhpharm-2015-000833 URL [本文引用: 1] 摘要
ObjectivesTo determine an effective framework for supplying emergency drugs under various scenarios using ‘modular design’ and information technology. Additionally, medicinal safety was improved by combining pharmacy monitoring with a safety alert system for medication.MethodsData from various emergency events and details of the disease related to the incident were analysed using Cluster, Delphi and Decision analyses. The optimal drug combination was determined and then divided into the different modules. We established the ‘drug supply expedited system in emergencies’ based on the above modules, and we organised emergency drills to verify the system9s effectiveness and to improve efficiency. Pharmaceutical care services were performed by rehearsing the unexpected emergency incident and associated pharmaceutical care.ResultsWe developed a drug supply framework for ‘traffic accidents, poisoning first aid, natural disasters, epidemics and mass disturbances’ and established an ‘emergency drug supply expedited system’. We quickly equipped the drugs that were needed for the special emergency events, and we developed a ‘green channel’ between the emergency and drug supply centres. Medication safety was also important for the emergencies, and clinical pharmacists played a role in medicating the safety service personnel. The utility of our findings was demonstrated through several emergency drills.ConclusionsIn this study, we explored the optimal drug supply and pharmacy assistance models for emergency medicine. The clinical innovation of this study was that we provided a modular supply of medical supplies for traffic accidents. We also established a drug supply information system. This study provided effective reference values for emergency drug therapy.
|
| [3] |
基于Apriori算法的数据挖掘在移动医疗系统中的应用 [J].Application of Data Mining in Mobile Health System Based on Apriori Algorithm [J]. |
| [4] |
数据挖掘技术在药品零售经营决策支持中的应用 [J].https://doi.org/10.3969/j.issn.1006-3765.2016.08.149 URL [本文引用: 1] 摘要
目的探究在医院药房药品零售经营决策支持中应用数据挖掘技术的效果及价值。方法从我院2013年01月起,在我院药房零售经营工作中建立微软SQL Server 2010数据工作库,并且运用关联规则法对我院药品日常销售产生的数据作数据挖掘和相关分析。结果顾客购入抗眩晕及维生素药物后,再购入理血药物的可能性大;同时购入解毒消炎药及清热解毒类药物饮片后,再购入五官科用药的几率大;购入心脑血管药物、消化系统药物的顾客主要为中老年患者,再购入理血药物的可能性大;夏秋季节清热解毒类中药饮片的销售额较高,顾客在购入咽喉含片时,有较大的可能性购入清热解毒类中药饮片;购入补益药、祛风湿药物之后,再购入骨伤科用药的几率较大。结论在医院药房药品零售经营决策支持中应用数据挖掘技术,便于分析和判断顾客的购物规律和趋势,能够为完善零售经营决策方案提供有价值的信息支持。
Application of Data Mining Technology in Decision Support of Drug Retail Business [J].https://doi.org/10.3969/j.issn.1006-3765.2016.08.149 URL [本文引用: 1] 摘要
目的探究在医院药房药品零售经营决策支持中应用数据挖掘技术的效果及价值。方法从我院2013年01月起,在我院药房零售经营工作中建立微软SQL Server 2010数据工作库,并且运用关联规则法对我院药品日常销售产生的数据作数据挖掘和相关分析。结果顾客购入抗眩晕及维生素药物后,再购入理血药物的可能性大;同时购入解毒消炎药及清热解毒类药物饮片后,再购入五官科用药的几率大;购入心脑血管药物、消化系统药物的顾客主要为中老年患者,再购入理血药物的可能性大;夏秋季节清热解毒类中药饮片的销售额较高,顾客在购入咽喉含片时,有较大的可能性购入清热解毒类中药饮片;购入补益药、祛风湿药物之后,再购入骨伤科用药的几率较大。结论在医院药房药品零售经营决策支持中应用数据挖掘技术,便于分析和判断顾客的购物规律和趋势,能够为完善零售经营决策方案提供有价值的信息支持。
|
| [5] |
聚类分析方法在建立自动化药房快速配药区中的应用 [J].Application of Cluster Analysis Method to the Establishment of Rapid Dispensing Area in Automated Pharmacy [J]. |
| [6] |
Using Link Analysis to Explore the Impact of the Physical Environment on Pharmacist Tasks [J].https://doi.org/10.1016/j.sapharm.2015.09.011 URL PMID: 26508715 [本文引用: 1] 摘要
National community pharmacy organizations have been redesigning pharmacies to better facilitate direct patient care. However, evidence suggests that changing the physical layout of a pharmacy prior to understanding how the environment impacts pharmacists' work may not achieve the desired benefits. This study describes an objective method to understanding how the physical layout of the pharmacy may affect how pharmacists perform tasks. Link analysis is a systems engineering method used to describe the influence of the physical environment on task completion. This study used a secondary data set of field notes collected from 9 h of direct observation in one mass-merchandise community pharmacy in the U.S. State, Wisconsin. A node is an individual location in the environment. A link is the movement between two nodes. Tasks were inventoried and task themes identified. The mean, minimum, and maximum number of links needed to complete each task were then determined and used to construct a link table. A link diagram is a graphical display showing the links in conjunction with the physical layout of the pharmacy. A total of 92 unique tasks were identified resulting in 221 links. Tasks were sorted into five themes: patient care activities, insurance issues, verifying prescriptions, filling prescriptions, and other. Insurance issues required the greatest number of links with a mean of 4.75. Verifying prescriptions and performing patient care were the most commonly performed tasks with 36 and 30 unique task occurrences, respectively. Link analysis provides an objective method for identifying how a pharmacist interacts with the physical environment to complete tasks. This method provides designers with useful information to target interventions to improve the effectiveness of pharmacist work. Analysis beyond link analysis should be considered for large scale system redesign.
|
| [7] |
Selecting a Pharmacy Layout Design Using a Weighted Scoring System [J].https://doi.org/10.2146/ajhp100687 URL PMID: 22517024 [本文引用: 1] 摘要
A weighted scoring system was used to select a pharmacy layout redesign. Facilities layout design techniques were applied at a local hospital pharmacy using a step-by-step design process. The process involved observing and analyzing the current situation, observing the current available space, completing activity flow charts of the pharmacy processes, completing communication and material relationship charts to detail which areas in the pharmacy were related to one another and how they were related, researching applications in other pharmacies or in scholarly works that could be beneficial, numerically defining space requirements for areas within the pharmacy, measuring the available space within the pharmacy, developing a set of preliminary designs, and modifying preliminary designs so they were all acceptable to the pharmacy staff. To select a final layout that could be implemented in the pharmacy, those layouts were compared via a weighted scoring system. The weighted aspect further allowed additional emphasis on categories based on their effect on pharmacy performance. The results produced a beneficial layout design as determined through simulated models of the pharmacy operation that more effectively allocated and strategically located space to improve transportation distances and materials handling, employee utilization, and ergonomics. Facilities layout designs for a hospital pharmacy were evaluated using a weighted scoring system to identify a design that was superior to both the current layout and alternative layouts in terms of feasibility, cost, patient safety, employee safety, flexibility, robustness, transportation distance, employee utilization, objective adherence, maintainability, usability, and environmental impact.
|
| [8] |
关于优化门诊药房工作流程的调查分析 [J].Analysis and Survey of the Optimization of the Workflow of Outpatient Pharmacy [J]. |
| [9] |
自动化药房系统应用于我院门诊药房的实践与体会 [J].https://doi.org/10.6039/j.issn.1001-0408.2016.19.24 URL [本文引用: 1] 摘要
目的:为我国医院现代化药房建设和发展提供参考经验,推动药房自动化系统在医院的应用。方法:通过介绍我院门诊药房自动化系统(快速发药机、智能存取机)的调试与应用情况及由此带来的药房管理的相应改变,提出我院自动化系统应用中存在的问题及相应的解决措施,并评价自动化系统在我院的应用效果。结果:配合自动化系统的使用,我院药学部调整了药房布局、对上机药品进行了调试优化、改进了加药和盘点操作、制订了门诊药房合理排班作业内容、建立了完整的药品有效期管理方案等;针对自动化系统出现的问题,制订了快速发药机和智能存取机规范的加药流程及相关作业书;针对自动化系统可能出现的故障制订了快速发药机加药、出药故障应急处理流程及智能存取机故障应急处理流程。经对相关指标进行评价,表明应用自动化系统并对其不断改进后,加药药师劳动强度降低(加药步行数由5 634.6步/d减少至4 087.8步/d),发药药师工作效率提高(高峰时段发药处方数由226.55张/h升高至311.55张/h),调配药师工作差错降低(调配内差数由54.75件/周减少至21.50件/周)。结论:门诊药房应用自动化系统后,规范了药房的调剂及人员管理等,已成为医院药学发展的必然趋势。但建议在应用过程中应对其不断进行调整及优化,以促进其发挥最大效能。
Practice and Experience of the Application of Automatic Pharmacy System in Outpatient Pharmacy of Our Hospital [J].https://doi.org/10.6039/j.issn.1001-0408.2016.19.24 URL [本文引用: 1] 摘要
目的:为我国医院现代化药房建设和发展提供参考经验,推动药房自动化系统在医院的应用。方法:通过介绍我院门诊药房自动化系统(快速发药机、智能存取机)的调试与应用情况及由此带来的药房管理的相应改变,提出我院自动化系统应用中存在的问题及相应的解决措施,并评价自动化系统在我院的应用效果。结果:配合自动化系统的使用,我院药学部调整了药房布局、对上机药品进行了调试优化、改进了加药和盘点操作、制订了门诊药房合理排班作业内容、建立了完整的药品有效期管理方案等;针对自动化系统出现的问题,制订了快速发药机和智能存取机规范的加药流程及相关作业书;针对自动化系统可能出现的故障制订了快速发药机加药、出药故障应急处理流程及智能存取机故障应急处理流程。经对相关指标进行评价,表明应用自动化系统并对其不断改进后,加药药师劳动强度降低(加药步行数由5 634.6步/d减少至4 087.8步/d),发药药师工作效率提高(高峰时段发药处方数由226.55张/h升高至311.55张/h),调配药师工作差错降低(调配内差数由54.75件/周减少至21.50件/周)。结论:门诊药房应用自动化系统后,规范了药房的调剂及人员管理等,已成为医院药学发展的必然趋势。但建议在应用过程中应对其不断进行调整及优化,以促进其发挥最大效能。
|
| [10] |
Satisfaction of Clients with the Services of an Outpatient Pharmacy at a University Hospital in Northwestern Ethiopia: A Cross-Sectional Study [J].https://doi.org/10.1186/s12913-015-0900-6 URL PMID: 26062912 [本文引用: 1] 摘要
Background Evaluation of patient/client satisfaction with pharmacy services as a crucial part of the health services through appropriate studies is important. This will help identify specific areas of the service which need improvement in realizing high quality pharmacy services in general and enhance the positive changes in the current pharmaceutical services provision in Ethiopia. The current study aimed at assessing the level of client satisfaction with the services of the outpatient pharmacy of Gondar University Referral Hospital (GURH) in northwestern Ethiopia. Methods An institution-based cross-sectional study was conducted involving 400 clients who had prescriptions/orders filled at the outpatient pharmacy of the hospital during the period of 5th to 25th of November 2013. The data on the level of satisfaction of clients with the services of the outpatient pharmacy in the hospital was collected using a structured interview guide adopted from an instrument translated into Amharic and validated. The data collected was entered into and analyzed using Statistical Packages for Social Sciences (SPSS) version 16. Results The overall mean score the respondents gave to satisfaction with the pharmaceutical services was 2.48 out of a maximum of 5.00 score. The mean scores for all the individual parameters rated were less than 3.00. Maximum mean scores were given for parameters asking about the promptness of prescription medication service (2.99), and professionalism of the pharmacy staff (2.96) with the lowest being scored for information given to clients about the storage of medication (1.25), and explanations of possible side effects (1.27). Clients who were served free of fee recorded significantly higher level of satisfaction than those who paid. Higher levels of satisfaction were also reported among illiterates, older adults and those with no job compared to those with higher education, merchants and government employees. Conclusions This study showed that the overall mean satisfaction level of clients of the outpatient pharmacy was low and it differed among different socio-demographic characteristics. Further research in to the reasons behind the low satisfaction should be done to provide appropriate solutions to improve the service.
|
| [11] |
Quality-Improvement Initiatives Focused on Enhancing Customer Service in the Outpatient Pharmacy [J].https://doi.org/10.2146/ajhp150152 URL PMID: 26272897 [本文引用: 1] 摘要
The development and implementation of quality-improvement initiatives to enhance customer service in an outpatient pharmacy of a Veterans Affairs (VA) medical center are described.Historically low customer service satisfaction rates with the outpatient pharmacy at the Philadelphia Veterans Affairs Medical Center prompted this quality-improvement project. A three-question survey was designed to be easily and quickly administered to veterans in the outpatient pharmacy waiting area. Using 5-point Likert scale, veterans were asked to rate (1) their overall experience with the outpatient pharmacy service and (2) their satisfaction with the customer service provided by the pharmacy department. They were also asked how they thought the pharmacy department could improve its customer service. After receiving feedback from the survey, several quality-improvement initiatives were developed. The initiatives were categorized as environmental, personnel, communicative, and technological. For each initiative, one or more tasks were developed and the initiatives were subsequently implemented over eight months. After each task was completed, veterans were surveyed to measure the impact of the change. A total of 79 veterans were surveyed before the implementation of the quality-improvement initiatives, and 49% and 68% rated their experience with the outpatient pharmacy and customer service favorably, respectively. Twenty-five veterans were surveyed after the implementation of numerous quality-improvement interventions, with 44% and 72% rating their experience with the outpatient pharmacy and customer service favorably.Customer service satisfaction with an outpatient pharmacy service at a VA medical center was enhanced through the implementation of various quality-improvement initiatives.
|
| [12] |
Building a Business Case for an Outpatient Pharmacy [J].
To be successful, an outpatient prescription pharmacy service built and operated by a hospital should be run as a competitive business, not in the manner of an inpatient operation. The outpatient pharmacy should not be a siloed entity that operates separately from the inpatient pharmacy. A hospital may miss a significant margin opportunity if it runs the pharmacy strictly as a safety net for uninsured or underinsured patients.
|
| [13] |
Removal of Duplicate Rules for Association Rule Mining from Multilevel Dataset [J].https://doi.org/10.1016/j.procs.2015.03.106 URL [本文引用: 1] 摘要
Association rules are one of the most researched areas of data mining. This is useful in the marketing and retailing strategies. Association mining is to retrieval of a set of attributes shared with a large number of objects in a given database. There are many potential application areas for association rule approach which include design, layout, and customer segregation and so on. The redundancy in association rules affects the quality of the information presented. The goal of redundancy elimination is to improve the quality and usefulness of the rules. Our work aims is to remove hierarchical duplicacy in multi-level, thus reducing the size of the rule set to improve the quality and usefulness without any loss.
|
| [14] |
Implementation of Association Rules with Apriori Algorithm for Increasing the Quality of Promotion [C]// |
| [15] |
一种带记忆性的零售商品关联度分析方法 [J].
基于关联规则的购物篮分析对实体或在线零售企业改善品类管理、优化门店布局陈列、实现关联促销与推荐等起着重要作用.本文针对企业需求,应用邻接矩阵和截矩阵技术,提出了挖掘商品关联性的AC算法和进一步求频繁k-项集的M-AC算法,证明了这类方法运算简单,应用成本低,具有记忆性质,当数据更新时,无须重新运算.通过在零售企业的成功应用,论证了这种商品关联度分析方法的有效性.
An Approach for Retail Goods Association Rules Analysis with Memory Property [J].
基于关联规则的购物篮分析对实体或在线零售企业改善品类管理、优化门店布局陈列、实现关联促销与推荐等起着重要作用.本文针对企业需求,应用邻接矩阵和截矩阵技术,提出了挖掘商品关联性的AC算法和进一步求频繁k-项集的M-AC算法,证明了这类方法运算简单,应用成本低,具有记忆性质,当数据更新时,无须重新运算.通过在零售企业的成功应用,论证了这种商品关联度分析方法的有效性.
|
| [16] |
|
| [17] |
BIRCH: An Efficient Data Clustering Method for Very Large Databases [C]// |
| [18] |
The Mating Pool: A Testbed for Experiments in the Evolution of Symbol Systems [C]// |
| [19] |
A Density Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise [C]// |
| [20] |
Improving Inference Through Conceptual Clustering [C]// |
| [21] |
|
| [22] |
|
| [23] |
Fast Algorithms for Mining Association Rules [C]// |
| [24] |
An Efficient Algorithm for Mining Association Rules in Large Databases [J]. |
| 版权所有 © 2015 《数据分析与知识发现》编辑部 地址：北京市海淀区中关村北四环西路33号 邮编：100190 电话/传真：(010)82626611-6626，82624938 E-mail:jishu@mail.las.ac.cn |
/
| 〈 |
|
〉 |
