The Research Of The Supporting System Of Medical Decision Based On The DW

Based on the discussion of the DSS and data warehouse technology, the paper adopt oriented-object technology to realize the Model Base design. And apply Data Mining techniques to discover the potential knowledge from lots of medical data. They are as follows: (1)Design concept model and logic model and acquire multi dimension data source for models. (2) Designing the class of model and completing the design of model base According to the class of model. (3) Three models are up built by association rule technique ,and a great deal of data of the hospital is mined . I. The Concept of Data Warehouse Data Warehouse is the key means of data storage and resources management based on databases, the effective way of data analysis and information taking by the statistic and analytic technology, the scientific way of knowledge exploring and laws discovery by the artificial intelligence technology, and the technology of the multi-sciences combination and the application. Data Warehouse can make good use of the present data resources to help the users understand the information better and treat them from a new aspect so that they can get better insights , discovery mode and trend to dig information and knowledge and make better assistant decision. II. The Organization and the Structure of Data Warehouse III. The Design of Data Warehouse The design of Data Warehouse can be made as follows: 1. The design of the conceptual model 2. The technological preparation 3. The design of the logical model 4. The design of the physical model 5. The generation of Data Warehouse 6. The using and the maintenance of Data Warehouse IV. Case 1. Defining the subjects: Subject is an abstract concept . It is the abstraction which combines, classifies and analyze the data from the enterprise information system in a higher level. The subjects in this case mainly include the following elements: 1) sufferer 2) disease 3) time 2. The definition of dimension It is used to make the angle to observe the data. 1) the disease dimensional table 2) the time dimensional table 3) the basic-situation dimensional table of the sufferer 4) the climate dimensional table 5) the popular-situation dimensional table of the other diseases 3. The Star-shape Model of the Medical Data Warehouse 4. Transformation The data from Data Warehouse which faces the subject is complex, summary and structural one, which cannot be got from the original operative database. Therefore, the original and distractive data from the database must be integrated and combined to the Data Warehouse. For example, the time dimension is a special one , which changes the date patterns from different data origins into the same pattern, and turns the values of each dimensional property into those that meet the need according to the level definition of the time dimension by users, then stores them into the corresponding record position of the property. For example, the date pattern of the data origin is 03/18/2005 or 2005/03/18 or the other pattern, while the pattern from the time dimension is year—quarter—month, then when the data is taken ,the date will be changed to: 2005,3,18. 5.The Data Load of the Medical Data Warehouse Data Load is to load the extracted and changed data from the data originalsystem to the system of the Data Warehouse, which needs the tools with highly effective loading property. In the supporting system of the medical decisions, the following methods are taken to load the data: 1) For the relational table of SQL Server which matches with the data structure, SQL insert is loaded, eg: insert into t_medinfo from oldt_medinfo 2) For the data from the different data origins which includes the relational database, document system , etc, the data load can be realized through DTS in Microsoft SQL Server 2000. 3) For the data to be arranged, it will be transformed to the fixed text document by the programme arrangement, and then be put into the Data Warehouse. 4) Loading the data by programming; 5) For a little amount of data, input them manually. 6.Clearness Data Clearness is the dispensable step of the design of Data Warehouse. On one hand, Data Warehouse is made for the decision supporting. The correct decision only can be made with the base of the complete and correct data. On the other hand, Data Warehouse is usually concerned with the great amount of data from several data origins. It is very possible that there's something wrong with the data. Therefore, the inspection and the correction of the unusual data is very valuable. V. The Functions of the Decision Supporting System Although it can provide the assistant decision for the decision-makers, the decision supporting system can't make decisions. In general, DSS has the following functions: 1.DSS mainly deals with the semi-structural and the non-structural problems in the decisions. The decision supporting is to achieve the following purposes through computers: (1)To help the decision-maker to make decisions in the semi-structural and the non-structural tasks. Structural Problems: the problems which can be described clearly. The definite calculation and the decision rules can be used in all the three stages. Non-structural problems: the problems which can't be described clearly, but only can be judged by tuition or experience. The definite calculation and the decision rules can Not be used in all the three stages. Semi-structural problems: The problems between the above two problems. The definite calculation and the decision rules can be used in one or two stages. The development of DSS can solve the semi-structural decision problems effectively and make the non-structural decision problems turn into the structural problems.(2)To support rather than replace the decision-makers'judgment and to improve the effectiveness of the decision rather than increasing its efficiency. 2. What DSS provides is to assist the decision-makers to make decisions rather than replace the decision-makers to make decisions. 3.DSS demands the greater interactive function between man and computer 4.DSS demands the ability of managing the mathematics model and a great amount of data. VI. The Basic Concept of MDSS The computer can assist the medical decision through two ways of the direct help and the indirect help. The indirect help system is to improve the speed and the quality of the medical decisions indirectly through the automation of the information processing. For example, the information system and the cases management system of the hospital can simplify the process of the cases obtaining (eg. inspection data) and the data analysis (report and the sufferer's table); Meanwhile, the data in the special fields(eg: document database, medicine or law database)can be gained through these systems. Direct help is to help the medical staff to make decisions to some sufferer directly. It applies the medical knowledge to the special problem of some sufferer and put foreword the best proportion between the effectiveness and the medical fees. The direct help system is the supporting system of the medical decisions. VII.The Structure of the Medical Decision Supporting System The article regards the knowledgebase as : fact+rule, if the fact is put into the database, the knowledgebase can be simplified to rulebase. Meanwhile, In order to reduce the complex of the sysem, the methodbase is put into modelbase, then the traditional four-base structure is simplified to the three-base structure (database, rulebase and modelbase), therefore, the functions of each part are independent and eliminate the redundancy of the system functions in some degree. MDSS is base on the kind of system.VIII.The Design of the Model The article builds three models, of whick Model_Apriori and Model_FP are both the exploring models of the single-dimension relational rule, but their working efficiencies are different. Model_FP works one time faster than Model_Apriori while Model_Multi is used to explore the multi-dimension relational rules. (1) Model_Apriori is uded to explore the knowledge from the present database so that the manager of the hospital can manage the hospital through the explored knowledge. The Aprior calculation of the relational rule is applied to this model. (2) The frequent-pattern growth model based on FP-Growth calculation. (3)Model_multi is based on the multi-dimension relational rule of the relational rules , which not only considers the relations between the sets of the item but also the confinement of the sets of the item. Model management Data management Model base Data Warehouse Sharing data interface Fig.2 MDSS based on the DW ELT of data metadata Inner,local outer,distance outer original data Rule base Interface of man-computer...