Research And Development Of The Auxiliary Medical Platform System Based On The Cloud Service

With the increasing standardization, marketization of cloud service applications,many other industries also begin to provide cloud services.in the medical industryabroad, cloud services have been a part of socialized life, providing mature cloudmedical applications, and making a balanceddistributionof medical resources, users canget medical information by a terminal intelligent equipment through the internet at anytime, which cannot be limited by time or region. While interiorly, the cloud services forthe medical field has just started, and has not reached the level of foreignstandardization and marketization, on the other hand, the current domestic cloudservices focus on establishing the medical records and information exchange, theservices of analyzing medical data didn’t have a greater development. Therefore, how tobuild up a distributed storage-based auxiliary medical platform system with the functionof analyzing data by data mining has become a hot point of present cloud medicalservices’ research.This paper’s work mainly includes the following four aspects:1、 Build up the distributed storage system and the distributed computingframework, using the distributed storage system to storage large-scale data, using thedistributed computing framework to execute coarse-grained calculation.2、Aiming at the efficiency issues caused by coarse-grained distributed computing,adding high performance computing to local nodes, making fine-grained results aboutnodes, thus solve the problem about the coarse distributed computing results, andaccelerate related calculation about data analysis. 3、Since the problem that clustering center of clustering partitioning algorithmK-Means is unstable, this paper explores a new clustering algorithm AIK-Means, whichuse hierarchical clustering algorithm Chameleon to improve the initial cluster centers ofK-Means algorithm, use the FP-Tree algorithm in correlation analysis to handle themultiple clustering results made by K-Means algorithm. Then applying the AIK-Meansalgorithm and Gaussian regression algorithm to distributed strategy in order to makeparallel execution and improve efficiency.4、For the access to multimode data in local nodes, this paper designs andimplements the URAM(Uniform Resource Access Middleware), and distributed systemcan achieve transparent data access to multi-mode.