透過您的圖書館登入
IP:18.232.187.47
  • 學位論文

經皮冠狀動脈氣球擴張術後再住院之評估研究

A Study on Postoperative Readmissions of Percutaneous Transluminal Coronary Angioplasty

指導教授 : 張俊郎
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


冠狀動脈疾病(Coronary Heart Disease, CAD)為國人常見的心臟疾病。根據行政院衛生署資料統計,民國101年心臟疾病在國人十大死因中高居第二,而經皮冠狀動脈氣球擴張術(Percutaneous Transluminal Coronary Angioplasty, PTCA)是治療冠狀動脈疾病相當常見的手術,國內接受PTCA治療之病例也逐年增加中。然而,當病患在接受經皮冠狀動脈氣球擴張術時,手術過程中可能傷害病患血管內皮細胞造成血管內皮細胞剝離,導致術後3-6個月內血管再狹窄發生機率高(約30-40%),造成病患必頇再次手術及再次住院觀察情形發生。因此,如何有效評估PTCA病患術後是否再住院,降低病患再住院機率,對醫師、病患、家屬以及醫療資源之利用及整體規劃皆有一定的重要性。 本研究以全民健保資料庫2010年接受PTCA之病患為研究對象,彙整影響PTCA術後再住院相關因子,運用倒傳遞類神經網路、貝氏網路、粒子群演算法、基因邏輯斯模式以及其相互結合之人工智慧方法,建構六個PTCA術後再住院之預測模型,再以倒傳遞類神經網路、粒子群演算法及基因邏輯斯模式之權重做為預設權重導入案例式推理系統,設計一再住院評估系統。研究結果顯示在預測模型方面,大部分模型皆有90%以上之平均測試準確率,其中又以倒傳遞類神經網路模型表現最佳,平均測試準確率達95.28%、平均ROC曲線下面積達0.838。而再住院評估系統方面,也是以倒傳遞類神經網路之權重做為預設權重結合案例式推理有著最佳表現,系統準確率達85.74%,平均相似度為97.97%。藉由本研究之結果可提供相關醫療人員、病患以及家屬做為術後保健之參考依據,有效地提升醫療與手術品質,同時避免醫療資源二次耗用之情況發生。

並列摘要


Coronary Artery Disease is a common heart disease in Taiwan. According to statistics of Department of Health, heart disease ranked as the second of the top ten causes of death in 2012. Percutaneous Transluminal Coronary Angioplasty (PTCA) is a fairly common treatment for coronary artery disease and cases of patients receiving such surgery are increasing yearly. However, patient’s vascular endothelial cells might be damaged during the procedure of PTCA which tends to result in endothelial cell detachment as well as the occurrence of restenosis within 3-6 months after surgery at a high probability of about 30-40%). Such condition leads patients to receive surgery again and to hospitalize for observation. Therefore, how to effectively evaluate whether patients after PTCA are required to re-hospitalize and to reduce patient readmission probability is of certain important for physicians, patients, families, and utilization and overall planning of healthcare resource. Based on data of patents undergone PTCA in the National Health Insurance database of 2010, this research compiled factors relating to t readmission associated to PTCA and applied back-propagation neural networks, Bayesian networks, particle swarm optimization, genetic logistic model as well as their mutual combination of artificial intelligence methods to construct six prediction models for PTCA postoperative readmission, and then to design a readmission evaluation system through inputting the default weights of back-propagation neural networks, genetic algorithm and particle swarm weights to Case-Based Reasoning (CBR). Results findings showed that, most prediction models were with high average accuracy rate of more than 90%, in which the back-propagation neural networks demonstrated best performance with the tested accuracy rate at 95.28%, and the average area under the ROC curve at 0.838. The readmission evaluation system also showed that the combination of the default weights of back propagation neural networks with CBR had the best performance, with the system accuracy rate of 85.74% and the average similarity of 97.97%. Thus, the findings of this study not only can serve as postoperative care reference basis for relevant medical staff, patients and their families, but also effectively improve qualities of medication and surgery as well as avoid secondary consumption of medical resources.

參考文獻


1. 三軍總醫院心臟內科衛教資訊,「冠心症」,取自http://wwwu.tsgh.ndmctsgh.edu.tw/cv/EDU/cv-3/cv-3.html,參考日期:2013/5/28。
2. 于博芮、劉向援、郭素娥、馮清淳、王佳芸,華杏編輯部(2010),成人內外科護理(五版),華杏出版公司,台北。
3. 王派洲、李慧慈、賴威利(2003),「使用混合基因迴歸法預測網路銀行使用者」,第八屆台灣企業個案研討會。
4. 王派洲(2008),資料探勘:概念與方法,滄海書局,台中。
5. 王庭荃、陳長興(2008),「醫師年資、醫療服務量與消化性潰瘍治療效果之相關研究」,台灣衛誌,27卷,1期,頁57-66。

延伸閱讀