Implementation Of Decision Support System For Medicare Designated Medical Institutions

In recent years, with the increasing of data scale, it is a great challenge to mine the information and knowledge from business history data about the designated medical institutions. This paper develops a decision support system to solve the problem in component-based software ideas.The intelligent decision support system is designed based on the data warehouse, data mining technologies and model bases. The logical structure of this system consists of three main aspects, including interactive components, data components and model components. In the B/S structure, decision-makers can easily use the decision support system with a browser. However, in the current frame of B/S technology, the EJB of J2EE is complex. Besides, the user experience of the application based on Struts2+Spring framework is ineffective. Taking these problems into account, this paper imports Flex technology to propose a combination framework of Flex, Struts2 and Spring framework. Further, a four-layer B/S framework is put up in this paper, including the view layer, control layer, business layer and persistence layer. This design not only makes developers devoting their efforts to specific business development, but also greatly improves the user experience by creating RIA (Rich Internet Applications).On the other hand, K-means algorithm is efficiently used in designated medical institutions unsupervised classification by quantitatively description of the designated medical institutions business and representing each of them as a point in a multi-dimensional space. However, the random selection of the initial value of K-means algorithm has a great influence on the clustering results. As a result, it tends to converge to a local optimal solution. Taking these occasions into account, this paper transforms the K-means clustering problem to the optimization problem of minimizing SSE (sum of squared errors). Further, this paper avoids the premature phenomenon by importing the particle swarm optimization (PSO), developing PSO based K-means clustering algorithm and combining PSO optimization algorithm global search capabilities. Further, two methods are proposed to optimize this algorithm, including reducing the w weight weights in the iterative process and adaptively altering the weights. Extensive experimental results show that our algorithm can converge to the global optimal solution effectively.Finally, this paper achieves the decision support systems of Medicare designated medical institutions based on the above designs. It greatly improves the data access speed by avoiding multi-table associated with the query according to the following two techniques, including modeling the medical establishment data with star structure and adding redundant fields in the fact table and encoded ID of medical institutions. Under the time series ARMA model and the PSO based K-means clustering algorithm, the forecast and cluster analysis of the designated medical institutions related businesses can help decision-makers to explore the potential information behind the data, gain valuable knowledge, and consequently provide decision supports for the labor and social security departments.