| Summary: | 碩士 === 稻江科技暨管理學院 === 網路系統學系碩士班 === 98 === In these years, knowledge discovery approach was applied in extracting intrinsic knowledge from the clinical medical process of handling and decision-making for ventilator weaning. These were clearly described with knowledge attributes in the contents of medical records. Data mining is widely utilized in various domains, and approaches have been proposed to extract medical information and mining rule, is one of the most important methods in artificial intelligence. Association analysis was adopt to implement medical knowledge mining in this study, 17 effective attributes were chosen as modeling variables in prior analysis. The attributes in the sample data of success case and failure case were filtered by means of CFS. 15 significant factors were screen out automatically and then apply to carry out the following analysis. For the success case, association rules were obtain with the condition of that min. support is 0.4 and min. confidence (lift) is 0.9, and these may summarize as 8 meaningful rules which only 11 attributes were dominant in these rules. In the verify results, only 3 examples in 101 male case completely meet the condition of success weaning, and does not have example in the failure case meet the condition of success weaning. So, it was unnecessary to coincide with all 11 indexes simultaneously in clinical application.
|
|---|
