結合FCA及CBR建構實證醫學混合知識本體應用於糖尿病人手術臨床決策支持系統

Integrating FCA and CBR to Construct an Evidence-Based Hybrid Ontology in a Clinical Decision Support System for Diabetic Patients Undergoing Surgery

鮑卓璨(Cho-tsan Bau)

指導教授 ： 陳榮靜

朝陽科技大學/資訊學院/資訊管理系/博士(2014年)

摘要

糖尿病會造成生活品質降低、多器官的併發症、和提早死亡，以及健康照護費用的增加。它是一個持續增加的重要健康問題，台灣地區超過一百萬的成人受到影響。由於糖尿病人接受手術的人數增加和糖尿病的相關併發症的風險升高，適當的手術評估和處置是非常重要的。
實證醫學是一門近年來盛行的醫學方法論。為使民眾有更好之醫療品質，國內外醫學專家學者致力於實證醫學的推動。由於知識管理的理念逐漸受到重視，本計畫乃提出利用本體論正規化概念(Formal Concept Analysis, FCA)分析建構CBR（Case-Based Reasoning）來支援實證醫學之知識工程方法論。希望藉由本計畫解析過去相似問題的解答來解決目前所面臨到的新問題，以期達到知識分享再利用之最終目的。
本研究進行的方式為，提出建構合適的醫學領域知識本體論的模式，將其應用在手術病人糖尿病處理實證醫學上。一方面提出手術病人糖尿病處理醫學領域知識本體論的整體架構，同時以糖尿病手術病人個案病歷檔案當基礎作為本計畫的知識範疇。從2008年3月至2010年11月共有307位糖尿病人於台中醫院接受手術治療，同時有402份的出院病歷儲存於案例資料庫。他們成功的治療經驗將被用來提升領域知識本體論的完整性。結合知識本體論(Ontology)及人工智慧技術等知識管理概念為基礎，以結構化、明確化有效的擷取臨床個案知識。達到糖尿病專家專業知識的分享和再利用，進而提升糖尿病手術病人的照護品質。
本研究使用史丹佛大學發展的Protégé來建構本體論及知識庫。運用FCA(Formal Concept Analysis)方法論來分析資料以產生臨床醫學的概念全文和概念圖。遵循Kuziemsky改良的四步驟來建立手術病人糖尿病處理醫學臨床本體論領域知識庫。以JENA規則語法來描述手術病人糖尿病處理的相關法則，然後透過JENA推論引擎來進行推論，提供臨床醫師手術病人血糖控制的建議。
根據上述的方法，我們建立一個臨床決策支援系統。包含一個手術病人糖尿病處理的知識本體，內有31個類、13個屬性；一個JENA推論引擎，含38條推論規則；一個模糊推論引擎，含27條模糊規則；以及一個友善的使用者介面。透過使用者介面輸入個案病人的資料，JENA推論引擎可以做出符合臨床路徑的決策建議。經由臨床醫療工作人員（3位糖尿病專家、3位主治醫師、3位住院醫師和3位專科護理師）的評估，知識本體的正確性和系統的有用性被確定，但是在滿意度和持續使用上，仍有改善的空間。因此，透過本研究，我們可以瞭解如何建構知識本體，同時與臨床決策支援系統做結合，用來協助臨床醫療工作人員。

關鍵字

模糊邏輯 ； JENA ； 手術 ； 糖尿病 ； 正規化概念分析 ； 本體論 ； 臨床決策支援系統

並列摘要

Diabetes mellitus (DM) causes a significant decrease in the quality of life, multisystem morbidities, and premature death. It is a global, significant health problem. The increasing prevalence of diabetics undergoing surgery and the risk of complications make optimal assessment and management during surgery extremely important. Evidence-Based Medicine (EBM) is proposed in order to provide good quality medicine. A clinical decision support system (CDSS) can effectively support physician performance and improve patient outcomes. The advantages of an ontology-based CDSS include sharing of clinical knowledge, update, and reuse. Evidence-based hybrid ontology for CDSS in the setting of diabetic patients undergoing surgery was created. We used a modified ontology development method to create the ontology and implemented the ontology using Protégé-OWL. From Mar 2008 to Nov 2010, 307 diabetic patients were admitted for surgery in Taichung Hospital and 402 serial discharge summaries were collected and stored in a database. The experiences detailed in these records can be retrieved and reused to improve the quality of medical treatment. Embedded clinical knowledge was elicited to complement the domain ontology with Case-Based Reasoning (CBR) and Formal Concept Analysis (FCA). The ontology included 31 classes and 13 properties, with 38 JENA rules and 27 Fuzzy rules built to generate recommendations. The evaluation studies confirmed the correctness of the ontology, acceptance of recommendations, and usefulness of the ontology for management of diabetic patients during surgery with the participation of 12 clinicians (3 endocrinologists, 3 attending physicians, 3 residents and 3 nurse practitioners. The contribution of this research is the setting up of evidence-based hybrid ontology and an evaluation method for CDSS. The primary experiment proves that the method is useful for clinical decision making.

並列關鍵字

Fuzzy Logic ； JENA ； Surgery ； Diabetes mellitus (DM) ； Clinical Decision Support System (CDSS) ； Ontology ； Formal Concept Analysis (FCA)

參考文獻

[2] Y. D. Jiang, C. H. Chang, T. Y. Tai, J. F. Chen, and L. M. Chuang(2012), "Incidence and prevalence rates of diabetes mellitus in Taiwan: Analysis of the 2000-2009 Nationwide Health Insurance database," Journal of the Formosan Medical Association, Vol. 111, No. 11, pp. 599-604.

[4] I. B. Hirsch, J. B. McGill, P. E. Cryer, and P. F. White(1991), "Perioperative management of surgical patients with diabetes mellitus," Anesthesiology, Vol. 74, No. 2, pp. 346-59.

[5] K. G. M. M. Alberti and D. J. B. Thomas(1979), "The management of diabetes during surgery," Br J Anaesth, Vol. 51, No. 7, pp. 693-710.

[6] L. A. Gavin(1992), "Perioperative management of the diabetic patient," Endocrinol Metab Clin North Am, Vol. 21, No. 2, pp. 457-75.

[8] H. L. Lazar, S. R. Chipkin, C. A. Fitzgerald, Y. Bao, H. Cabral, and C. S. Apstein(2004), "Tight glycemic control in diabetic coronary artery bypass graft patients improves perioperative outcomes and decreases recurrent ischemic events," Circulation, Vol. 109, No. 12, pp. 1497-502.