序列模式挖掘在临床路径发现问题中的研究与应用

发布时间：2018-01-10 23:28

本文关键词：序列模式挖掘在临床路径发现问题中的研究与应用　出处：《中国科学技术大学》2017年硕士论文　论文类型：学位论文

【摘要】：提高医院管理的效率和增加患者就医的透明度,规范医护人员的临床行为和减少医疗资源的浪费关乎每个人的切身利益,这是当前全社会关注的热点问题。临床路径作为一种临床医疗行为管理的手段,已经被世界上很多国家用于医院管理,并取得了很好的效果。我国从2009年开始进行临床路径的试点和推广工作,力争2020年底实现全部二级以上医院纳入临床路径管理。目前临床路径的制定是由相关领域医学专家根据经验和专业知识制定,费时费力。医疗信息技术的发展使得医院积累了海量的临床数据,促进了大数据技术在临床医学上的研究和应用,这为临床路径的发现和制定提供了新思路。数据挖掘技术能充分利用已有数据的信息,为临床路径的制定提供参考和指导,使临床路径的制定更具科学性和合理性。本文基于频繁序列模式挖掘对临床路径发现问题进行了研究。主要工作内容如下:(1)提出了有前缀约束的频繁序列模式挖掘算法CPM-PC(Clinical Pathways Mining with Prefix Constraints)。临床路径是针对某个病种的一套标准化的治疗方案,实际上是一系列检查、治疗以及护理行为按照时间先后顺序组成的一个序列,因此临床路径的挖掘问题被转化为频繁序列模式挖掘问题。另外,患者一个完整的治疗疗程总是由一些特定的医疗行为开始,像手术过程一般是从麻醉开始,据此在论文中提出临床路径"前缀集"的概念,即可以作为某种疾病治疗开端的临床项目的集合。在此基础上,提出了一个适用于临床路径发现的序列模式挖掘算法CPM-PC,该算法充分考虑临床路径"前缀集"的作用,能够较高效的挖掘出有意义的序列模式。(2)提出了临床路径基于属性的评估模型。由于临床路径的挖掘会产生大量的候选序列模式,对这些模式如何选择,目前还没有相关研究。对临床路径评估的研究大都集中于对已经试点的临床路径进行结果分析式的评估,这种方法不能适用于临床路径的选择过程。因此,在总结国内外临床路径评估方法的基础上,对各评价指标进行分析,选择住院时长LOS,医疗总费用C,药占比P三个基本属性提出了基于属性的临床路径评估模型,更加侧重于临床路径本身的属性,通过计算临床路径基于属性的加权评估值ABWE评价临床路径的优劣。最后,在阜阳市某医院的真实临床数据集上进行了实验,对实验结果的分析显示CPM-PC算法在挖掘临床路径过程中相比于传统的序列模式挖掘算法有更好的性能。
[Abstract]:To improve the efficiency of hospital management and increase patient transparency, standardize medical staff of the clinical behavior and to reduce the waste of medical resources related to the vital interests of each person, this is the current hot issues of concern to the whole society. Clinical pathway as a clinical medical behavior management means, has been in many countries in the world for hospital management, and good results have been achieved. China began to carry out the clinical pathway from 2009 pilot and promotion work, and strive to achieve by the end of 2020 all two hospitals in clinical pathway management. At present the development of clinical pathways is the domain of medical experts according to experience and expertise to develop, time-consuming and laborious. The development of medical information technology allows the accumulation of hospital the clinical data, promote the research and application of big data technology in clinical medicine, the clinical path discovery and formulation for New ideas. Data mining technology can make full use of the existing data information, to provide reference and guidance for the development of clinical pathway, make clinical pathway is more scientific and reasonable. The frequent sequential pattern mining based on the clinical path finding problem is studied. The main work contents are as follows: (1) proposed CPM-PC frequent sequence pattern mining (Clinical prefix constraint Pathways Mining with Prefix Constraints). Clinical pathway is a treatment for a disease of a standard, is actually a series of examinations, treatment and nursing behavior according to a sequence of time sequence composition, so the problem of clinical pathway mining mining was transformed into frequent sequence pattern. In addition, patients with a complete course of treatment is started by some specific medical behavior, as the operation process is generally from the beginning of anesthesia, The concept of clinical pathway "prefix set" in the paper, which is a collection of items can be used as a clinical disease treatment beginning. Based on the algorithm of CPM-PC sequence model proposed a suitable found in clinical pathway mining, the algorithm takes into account the clinical path "prefix set", can be more efficient the mining sequence patterns meaningful. (2) proposed the clinical pathway based on attribute evaluation model. Due to mining of clinical path will produce a large number of candidate sequence patterns, how to choose the mode, there is no relevant research. The study of clinical pathway evaluation mostly focuses on clinical pathway for the pilot has the analysis of the evaluation of the selection process, this approach cannot be applied to the clinical pathway. Therefore, based on summarizing the clinical pathway evaluation methods, the evaluation indexes were analyzed and selected Choose the length of stay in LOS, the total medical costs of C, the ratio of drug P three basic properties of proposed evaluation model of clinical pathway based on attribute, attribute more focused on the clinical pathway itself, by calculating the value of the advantages and disadvantages of ABWE clinical pathway evaluation of clinical pathway based on weighted assessment attributes. Finally, experiments are carried out in a hospital Fuyang City, the real clinical data sets, the analysis of the experimental results show that CPM-PC algorithm in mining sequential patterns of clinical pathway process compared to the traditional mining algorithm has a better performance.

【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13

【参考文献】