應用資料探勘技術對肩關節攝影檢查鑑別診斷差異性之探討-以某區域教學醫院為例

A Study on the Differential Diagnosis of Shoulder Arthrogram with the Application of Data Mining - Taking a Regional Hospital for an Example

謝宏其(Hsieh hung chi)

指導教授 ： 黃維民

國立中正大學/管理學院/資訊管理學系/碩士(2014年)

摘要

我國於1995年實施「全民健康保險」，此制度對於民眾之醫療需求、醫療與就醫行為、醫界生態、社會經濟，以至國家整體發展均有深遠的影響。為了改善健保的財務狀況，健保署積極著手改革支付制度，包括全面實施總額支付制度、擴大實施論病例計酬、合理調整支付標準等政策，以減輕全民健保的財務負擔。但是這些政策的施行除了限制了醫院的發展也衍生出許多的問題。
在2002開始實施總額的制度之下能夠使用的額度受到限制，所以在臨床執行醫療業務上也會產生排擠效應，如何讓檢查更有效的運用，使得有限的額度能照顧到更多需要的病人將是各個醫院努力的目標。
本研究個案醫院在骨科門診的病人量在平均每個月的骨科病患約在1500人至2000人之間。而至放射線科進行肩關節攝影之病患平均每個月約有20至25個病患數。每年約有240至300名病患，本研究統計97~100年之資料，研究病患數約在1000人左右。
本研究在資料分析的部份分為二部份，第一部分是檢查與鑑別診斷之準確性；第二部份則為利用邏輯迴歸分析的技術，採用資料關連 (data association)的模型，進行資料的分析來建立預測之模式。第一部分依醫院評鑑檢查之陰性率將閾值設定在50％，在鑑別診斷準確性的研究得到結果一致性為57.7％，達到所設定之閾值，而伴隨出現次數最多的則是與脊椎相關疾病之診斷碼。在第二部份邏輯迴歸分析，經由分析結果找出關聯性最大的診斷碼分別是715.9和727.61與715.9和840.4二組，研究分析所得到的結果與臨床醫師在臨床的診斷的邏輯是一致的。

關鍵字：關節攝影、邏輯迴歸分析、診斷碼

關鍵字

診斷碼 ； 邏輯迴歸分析 ； 關節攝影

並列摘要

Our implementation of the "National Health Insurance" in 1995, the medical needs of the people of this system for medical and health seeking behavior, the medical community ecology, socio-economic, as well as the overall development of the country have a profound impact. In order to improve the financial situation of health insurance, health care payment system department actively engaged in the reform, including the full implementation of the total payment system, expand the implementation of case payment, reasonable adjustments to payment standards and other policies, in order to reduce the financial burden on national health insurance. But the implementation of these policies in addition to limiting the development of the hospital also led to many problems.
Under the 2002 regime began to use the total amount is limited, so the clinical implementation of medical business will generate crowding out effect, how to make more effective use of checks, making a limited amount of energy needed to take care of the patient will be more is the goal of each hospital efforts.
The case study of hospital patients in outpatient orthopedic orthopedic patients in the average monthly amount of between about 1,500 to 2,000 people. And to make the shoulder of arthrography radiology patients about the average monthly number of patients 20-25. About 240-300 patients per year in this study statistical data 97 to 100 years, the number of patients studied in about 1,000 people.In this study, part of the data analysis is divided into two parts, the first part is to check the accuracy and differential diagnosis; second part was analyzed using logistic regression techniques, the use of information related (data association) model for data analysis to establish the prediction model. According to the first part of the examination of the negative rate of hospital accreditation threshold set at 50%, in the differential diagnostic accuracy studies to get consistent results for 57.7%, to reach the threshold of the set, and accompanied by the highest number is associated with spinal disease the diagnostic codes.In the second part of the logistic regression analysis to identify the most relevant results are analyzed by the diagnostic code is 727.61 and 715.9 and 715.9 and 840.4 the two groups, with the results obtained by analysis clinicians in clinical diagnostic logic is consistent .

Keywords: Arthrography, logistic regression analysis, diagnostic codes

並列關鍵字

Diagnostic codes ； Logistic regression analysis ； Arthrography

參考文獻

12. 陳垂呈（2007），利用資料探勘技術輔助疾病診斷是否異常，南台科技大學資訊管理系。

15. 姚志成（2004），運用資料探勘技術建構脂肪肝預測模式，中原大學資管系碩士論文。

21. 陳彥良等（2001），"資料間隱含關係的挖掘與展望," 二十一世紀台灣

35. 童俊榮（2012），應用資料採礦技術分析台灣健保資料庫-以攝護腺癌病人為例

36. 呂駿南（2013），應用資料探勘技術於十大慢性疾病關聯性之研究。