基于Petri网行为轮廓的医疗业务流程挖掘方法

发布时间：2018-07-10 19:57

本文选题：医疗流程挖掘 + 事件日志　；参考：《安徽理工大学》2017年硕士论文

【摘要】：许多医疗机构为了能提高其内部管理水平和提升对外服务的程度,打破原有管理的思维惯性,通过挖掘实际医疗业务流程并将其业务流程进行交互以此来达到优化系统,提高工作效率的目的。由于挖掘和交互容易受到内部和外部因素的干扰,导致挖掘和交互的模型跟实际医疗业务流程相悖,影响医疗业务系统的工作效率,增加工作量。因此,研究医疗业务流程挖掘方法并分析合并后的医疗流程的变化域具有很大的实际价值和理论意义。传统的分析方法大多以关联规则、聚类、决策树、神经网络和支持向量机等方法来分析相关数据,有些方法是由活动之间直接依赖的因果关系建立系统相关的业务流程模型。有些方法可以挖掘出活动之间的间接依赖关系但没有从根本上解决活动与活动之间的行为关系。在实际运行过程中常常导致死锁或虹吸情况。本文根据Petri网行为轮廓的理论及性质,提出了基于行为轮廓的医疗业务流程的挖掘和交互方法,以及医疗业务流程交互后产生的变化域分析方法。本文主要内容如下:(1)针对挖掘医疗系统中的实际业务流程问题,已有的研究都是建立在分析事件日志中活动之间的结构来挖掘流程,挖掘出的流程具有一定的缺陷性。本文利用Petri网行为轮廓的序关系确定事件日志中活动间的行为关系,提出了基于事件日志的医疗业务流程挖掘方法,从活动间的行为关系挖掘流程模型,并通过验证挖掘模型与实际流程的服从性确定挖掘模型的有效性。该方法提高了挖掘模型与实际事件日志的服从度。(2)针对不同医疗业务流程交互的方法以及交互后存在变化域的问题,已有的研究都是基于结构角度考虑的,通过分析流程中活动发生的先后顺序,进行简单的交互后忽略了交互后存在变化域问题,存在一定的局限性。本文利用Petri网行为轮廓关系确定流程间活动的匹配关系并确定最大匹配域,提出了基于行为轮廓的医疗流程模型合并及寻找变化域的方法,并分析产生变化域的原因。该方法结合模型中活动与活动之间行为的关系,分析模型合并交互后的变化域,提高了分析变化域的准确性。
[Abstract]:In order to improve the level of internal management and improve the degree of external service, many medical institutions break the thinking inertia of the original management, by mining the actual medical business processes and interacting their business processes to achieve the optimization of the system. The purpose of improving work efficiency. Because mining and interaction are easily interfered by internal and external factors, the model of mining and interaction is inconsistent with the actual medical business process, which affects the efficiency of medical business system and increases the workload. Therefore, it is of great practical value and theoretical significance to study the method of medical business process mining and analyze the changing region of medical process after merging. Most of the traditional analysis methods use association rules, clustering, decision tree, neural network and support vector machine to analyze the relevant data. Some of the methods are based on the causality of the direct dependence between the activities to establish a system-related business process model. Some methods can excavate the indirect dependence relationship between activities but not fundamentally solve the behavior relationship between activities. Deadlock or siphon situations are often caused during actual operation. According to the theory and properties of behavioral contour of Petri nets, this paper presents a method of mining and interacting medical business processes based on behavioral contour, and the analysis method of the change domain of medical business processes after the interaction of medical business processes. The main contents of this paper are as follows: (1) aiming at the actual business process problem in the medical system, the existing research is based on the structure between the activities in the analysis event log to mine the process. In this paper, we use the order relation of Petri net behavior profile to determine the behavior relation among the activities in the event log, and propose a medical business process mining method based on the event log, and mine the process model from the behavior relation between the activities. The validity of the mining model is determined by verifying the compliance of the mining model and the actual process. This method improves the serviceability of mining model and actual event log. (2) aiming at the problems of different medical business process interaction methods and the problem of the change domain after the interaction, the existing research is based on the structure perspective. By analyzing the sequence of the activities in the process, the problem of the domain of change is ignored after the simple interaction, and there are some limitations. In this paper, the behavior contour relation of Petri net is used to determine the matching relation between processes and the maximum matching domain. A method of merging the medical process model based on behavioral contour and finding the change domain is proposed, and the reasons for the change domain are analyzed. This method combines the relationship between the activity and the behavior in the model, and analyzes the change domain after the model is merged, which improves the accuracy of the analysis of the change domain.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R197.3;TP301.1

【相似文献】