基于云服务的辅助医疗平台系统的研究与开发

发布时间：2018-05-21 03:31

本文选题：APCM + 分布式系统　；参考：《西安建筑科技大学》2014年硕士论文

【摘要】：随着云服务应用的越来越规范化、市场化和标准化，其他行业也随之提供相应的云服务。在国外医疗行业，云服务已经进入社会化生活，提供了成熟的云医疗应用，使得医疗资源均衡分布，用户可以不受时间、地域的限制，通过网络利用终端智能设备随时随地获取医疗信息。在国内，将云服务用于医疗领域刚刚起步，并没有达到国外规范化和市场化的程度，并且，国内的云医疗服务目前重点在于建立病历档案和信息交互上，对医疗数据进行数据分析的云服务并未有较大发展。因此，如何建立一种以分布式存储为基础、以数据挖掘的方法分析数据为功能的辅助医疗平台系统，已成为当前云医疗服务中的一个研究热点。本文的主要工作是搭建云辅助医疗平台APCM（Assistant Platform forCloud-Medical），具体工作包括以下四个方面。 1、搭建分布式存储系统和分布式计算框架，利用分布式存储系统存储大规模数据并使用分布式计算框架执行粗粒度计算。 2、针对分布式计算的粗粒度计算所引起的效率问题，在本地节点中加入高性能计算，将节点中的计算进一步细粒度化，解决分布式计算过于粗糙的问题，，并加速了数据分析的相关计算。 3、针对聚类划分算法K-Means聚类中心不稳定的问题，探索了一种新的聚类算法AIK-Means，该算法利用层次聚类算法Chameleon改进K-Means算法的初始聚类中心，并且将关联分析中的FP-Tree树算法用于处理K-Means算法的多次聚类结果，从而找到频繁项集为最终结果。将AIK-Means算法和高斯回归算法基于分布式策略，使之并行执行，提高效率。 4、针对本地节点异构数据的访问问题，设计并实现了统一资源访问中间件URAM（Uniform Resource Access Middleware），分布式系统可以对异构数据实现透明访问。
[Abstract]:As the application of cloud services becomes more and more standardized, marketization and standardization, other industries also provide corresponding cloud services. In the foreign medical industry, cloud service has entered into social life, providing mature cloud medical applications, so that medical resources can be evenly distributed, and users can be free from time and geographical constraints. Access to medical information anytime and anywhere through the use of terminal intelligent devices through the network. At home, cloud services are just beginning to be used in the medical field, and they have not reached the level of standardization and marketization abroad. Moreover, the current focus of domestic cloud medical services is on establishing medical records and information exchange. Cloud services for data analysis of medical data have not developed significantly. Therefore, how to establish an auxiliary medical platform system based on distributed storage and analysis of data by data mining has become a research hotspot in cloud medical service. The main work of this paper is to build APCM(Assistant Platform for Cloud-Medical platform, which includes four aspects. 1. The distributed storage system and the distributed computing framework are built. The distributed storage system is used to store large-scale data and the distributed computing framework is used to perform coarse-grained computing. 2. Aiming at the efficiency problem caused by coarse-grained computing in distributed computing, high performance computing is added to the local node to further fine-grained the computation in the node, so as to solve the problem that the distributed computing is too rough. It also accelerates the calculation of data analysis. 3. Aiming at the instability of clustering center in K-Means, a new clustering algorithm, AIK-Means, is explored. The hierarchical clustering algorithm (Chameleon) is used to improve the initial clustering center of K-Means algorithm. The FP-Tree tree algorithm in association analysis is used to deal with the multiple clustering results of the K-Means algorithm, and the frequent itemsets are found as the final results. The AIK-Means algorithm and the Gao Si regression algorithm are implemented in parallel based on the distributed strategy to improve the efficiency. 4. Aiming at the problem of heterogeneous data access of local nodes, a unified resource access middleware (URAM(Uniform Resource Access Middleware) is designed and implemented. The distributed system can access heterogeneous data transparently.
【学位授予单位】：西安建筑科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09;TP311.52

【参考文献】