人口結構的改變使得全球朝向高齡化社會發展,國人於健康意識上亦逐漸抬頭,進而影響醫療機構的角色轉變,健檢的項目或服務為各大醫療機構積極耕耘的新藍海,本研究是以藍海策略探討應用於醫療健康產業之健康檢查中心,並以長春藤預防醫學健康管理中心為例,透過藍海策略的架構與工具進行應用分析。採用民國93~97年度「國民營養健康狀況變遷調查」計畫之體檢報告共有3,671筆樣本資料,排除1,827筆遺漏值,共採用1,844筆屬性資料,並將心血管疾病確診病例分為七種組合,運用Weka 軟體中C4.5(J48 in weka)、NB Tree、貝式網路(Bayes Net)、純樸貝式法(Naïve Bayes)及多層感知機(Multi-Layer Perceptron, MLP)之五種分類演算法進行資料分析,並利用混淆矩陣做分類分析比較,以探討分類準確度之變化,測試發現第一組(Group1)在全樣本訓練集及十折交叉驗證中之準確度皆較合理,在全樣本訓練集中以J48分類演算法精確度(Accuracy)為最高達81%;十折交叉驗證中以貝式網路(Bayes Net)之分類演算法表現較佳。進一步地,採用模糊理論建置心血管疾病之風險預測系統,提供使用者簡易之介面系統可便於評估個人罹患心血管疾病的風險程度,以作為國人自我健康管理或醫師於診斷疾病之參考工具。
Recently years, the change in the global population structure makes the nation toward an aged society. As a result, people increasingly concerns about their health such that the role of medical institutions also changes. The items or services of health examination become the new blue ocean for medical institutions. Accordingly, this thesis focuses on the study of Blue Ocean Strategy on the health evaluation center and takes the I-care Health Center as an example to proceed to the qualitative analysis under the framework of Blue Ocean Strategy. The dataset of medical examination, conducted by Nutrition and Health Survey in Taiwan from 2004 to 2008, is used in this work by excluding 1,827 samples and adopting 1,844 samples from 3,671 samples for the data mining analysis. Among the available samples, confirmed cases of cardiovascular disease are combined to 7 groups and analyzed using J48 algorithm, NB tree, Naïve Bayes, Bayes Net, and Multi-Layer Perceptron, respectively, which all have been implemented in the Weka software, and the accuracy of classification analyzed results are compared using confusion matrix. Experiments shown that the results of Group1 are acceptable when using all samples as the training set and ten-fold cross-validation. The best accuracy for the training set obtained using J48 is 81%, on the other hand, the performance of confusion matrix using ten-fold cross-validation is the best based on Bayes Net algorithm. Furthermore, this thesis adopts a fuzzy expert system to provide a personal evaluation interface for assessing the risk degree of cardiovascular disease and as a reference tool for self-health management and physician’s disease diagnosis.