Study On Data Warehouse-Based Decision Supporting System Of Hospital Management

Background: Along with China entering into WTO, the outside environment of hospital is changing rapidly, and bring forward higher require of the hospital decision making. With this trend and transformation, we need new tools and technology urgently to make decision for hospital, but currently hospital information system or on-line transaction processing doesn't have so strong function. Decision supporting system, which based on data warehouse, on-line analysis processing and data mining, can partly solve the problems above.Purpose: Through constructing data warehouse with multiple dimensionalities and minute granularities, making flexible on-line analysis processing and building decision tree and clustering data mining, can set up a decision supporting system which can be used for multidimensional decision analysis and for deep-seated data utilize.Methods: The study made half-structure interview for some hospital administrator, combining the literature analysis, confirmed the decision subject, using the Microsoft SQL server Analysis service as the tool of data warehouse, on-line transaction analysis and data mining. The data of the research roots in the historical database of a hospital information system, which related to fee information, clinical information, and the handbook of data structure.Results:⑴The decision subject of the research is the analysis of fee and clinic quality of in-patient.⑵Makes particularly design of the physical structure and logical model of the hospital management data warehouse database.⑶Building the fee analysis of in-patient multi-dimension data warehouse with time, disease, sex, age, identity, office, doctor, charge type, admission times, admission condition, patient class, operation condition, discharge disposition, treat result, fee sect, etc. about fifteen dimensions, and average treat days, average treat days before operation, average total fee, average drug fee, etc. about thirty-three measurements; building the clinic quality analysis of in-patient multi-dimension data warehouse with time, disease, sex, age, diagnose comparison group, diagnose correspondence, etc. about nine dimensions, and the rate of diagnose correspondence, the rate of recovery, the rate of healing, the rate of death, etc. about twelve measurements; particularly describes the meta data of the above two multi-dimension data warehouse.⑷Connecting the SQL server 2000 Analysis service server with Microsoft Excel Pivot Table, making on-line analysis processing from the basic condition of in-patient, fee composing, fee impacting factors and clinic quality analysis four aspect.⑸Constructing in-patient fee impacting factors decision tree data mining model and in-patient grouping clustering data mining model. In-patient fee impacting factors decision tree data mining model analyzes nine impacting factors and the intensity of the relevance. In-patient grouping clustering data mining model arranges all in-patients into three groups with obvious features, and analyzes the properties of each group.Conclusion:⑴The decision supporting system that the study constructs is combining the data warehouse, on-line analysis processing and data mining together. The three rely mutually, replenish mutually, and can display each one's superiority and achieving more efficient decision-making.⑵The dimensions and measurements of the data warehouse are designed according the guidebook of hospital management evaluating of the health ministry in 2005 and the analysis report of the third investigation of National Health Service in 2003,these make sure the scientific and practical of the statistic index. The disease dimension is designed according ICD-10, it contains ICD-10 main category, ICD-10 category, and ICD-10 subcategory. The patient dimension contains the treating office, the treating doctor and the particular patient.⑶The function of this decision supporting system is clear and definite, the dimensions of it is multiple, the granularities are minute, and the data showing is flexible, it can satisfies the demands of hospital administrators'decision-making from in-patient fee and in-patient clinic quality two sides. The fee analysis of in-patient multi-dimension data warehouse can analyze the average fee and fee composing from the whole hospital, every office and every doctor, making the hospital administrators can control the fee directed against; it also can analyze the changing tendency of fee, founding the cause of increasing or decreasing of the hospital's income. The clinic quality analysis of in-patient multi-dimension data warehouse can analyze the composing of the in-patients, can inquire about clinic quality index of different time, office and disease, and make latent tendency analysis of them, these will do great help to the analysis of the clinic quality and the natural law of disease, then the hospital administrators can take management measures, improving the efficiency and quality of the hospital.