任何意外的發生，對患者、家屬及相關醫療人員而言最重要的就是患者存活率的提升。然而在燒燙傷方面，因為礙於醫院燒燙傷中心病床床位等資源的有限，因此要如何於治療過程中做出有效的配置將有相當程度的影響。因此如果能在患者入院初期即可預先得知，何種燒燙傷情況的患者的死亡率會較高，將可以有助相關人員做出較佳的因應對策。
在發展燒燙傷患者死亡率的預測模型中，一般多以統計模型為主，當中又以羅吉斯迴歸分析及區別分析最常被採用。然而由於在使用一般的統計分析方法之前，需先確定分析資料有服從分析方法的統計基本假設，而且也只有在服從統計假設的前提下，得到的分析結果才有意義。
然而在實際狀況中所取得的資料，卻可能發生與統計假設相左右的情況，如此即造成資料在分析時易產生預測的偏差或無法取得隠匿的資訊。有鑑於此，本研究改採用一種無母數統計分析方法－約略集合理論進行分析，期望可以發展出一套較佳的燒燙傷死亡率預測模型。
而從本研究的實證結果中，透過對整體分析結果的比較可得知，約略集合理論是具有較佳表現的分析方法。同時經由約略集合理論分析所得到患者因燒燙傷致死的結果符合一般的認知，因此約略集合理論應可以適用於此一類形的研究領域。

While accident occurs, the most important thing for patients, family members, and medical personnel is to increase the patients’ survival rate. Since the sickbeds in the burned department of the hospital are limited, to arrange the sickbeds efficiently during the treatment process highly influence the survival rate. If medical personnel have knowledge in what types of burns causing higher mortality rate, they could make a better decision for patients.
In predicting the mortality rate, researches tend to use statistical models such as the logistic regression analysis and discriminant analysis. However, the data must fit the statistical assumptions in order to use these methods.
In real life, data are not always fit in the assumption and this will cause inaccuracy in predicting or unable to find out some important information. Therefore, we turn our attention to a nonparametric method - Rough Set Theory. We hope to come up a better prediction model for burn patients’ mortality rate.
From the empirical results, it shows the rough set theory performs well in some ways compared to other methods. Through the rules generated from the rough set theory, they conclude the same results of patients’ death from the common cognition. Thus, we believe that the rough set theory can be successfully applied into the same research fields.

摘要 iv
英文摘要 v
目次 vi
表目錄 viii
圖目錄 x
第一章 緒論 1
1.1 研究背景與動機 1
1.2 研究目的 2
1.3 論文架構 3
第二章 文獻探討 4
2.1 燒燙傷預測之相關研究及發展 4
2.2 資料分析方法 5
2.2.1 羅吉斯迴歸分析 5
2.2.2 區別分析 12
　　　　2.2.3 文獻小結 18
第三章 研究方法 19
3.1 約略集合理論的發展及應用 19
3.2 約略集合理論的操作流程及概念 19
3.3.1 資訊表 20
3.3.2 不可區分關係 22
3.3.3 近似集 25
3.3.4 屬性刪減與核心 30
3.3.5 決策規則 33
第四章 實驗評做與結果 36
4.1 資料說明與結構 36
4.1.1 資料說明 36
4.1.2 資料結構 38
4.2 不同狀態患者在條件屬性之分析 41
4.2.1 不同狀態患者在性別上之分析 41
4.2.2 不同狀態患者在年齡三等分上之分析 41
4.2.3 不同狀態患者在年齡十等分上之分析 42
4.2.4 不同狀態患者在住院週數上之分析 43
4.2.5 不同狀態患者在TBSA上之分析 43
4.2.6 不同狀態患者在吸入性燒燙傷上之分析 44
4.3 實驗步驟流程及分析結果 45
4.3.1 區別分析 46
4.3.2 羅吉斯迴歸分析 48
4.3.3 約略集合理論 50
4.4 分析結果比較 57
第五章 研究結論與建議 58
5.1 結論 58
5.2 研究限制與建議 61
參考文獻 62

1王保進，多變量分析（初版），高等教育文化事業有限公司，民93年。
2行政院衛生署，台灣地區歷年事故傷害死亡率，民國92年臺灣地區死因統計結果摘要。　　　　　　　　　　　　　　　　　　　　　　http://www.doh.gov.tw/statistic/data/死因摘要/92年/歷年事故傷害柱狀圖.xls。
3袁繼銓，以類神經網路預測燒傷病患住院日之研究，國立中山大學資訊管理研究所，民91年。
4陳順宇，多變量分析（二版），華泰書局，民89年。
5陳順宇，迴歸分析（三版），華泰書局，民89年。
6陳耀茂，多變量分析導論，全威圖書，民91年。
7黃俊英，多變量分析（七版），中國經濟企業研究所，民89年。
8黃登源，應用迴歸分析（初版），華泰文化事業股份有限公司，民87年。
9劉應興，非線性迴歸與相關分析，華泰書局，民87年。
10Bajgier, S. M., & Hill, A. V., “An Experimental Comparison of Statistical and Linear Programming Approaches to the Discriminant Problem”, Decision Sciences, Vol. 13, 1982, pp. 604-618.
11Bhatia, A. S. and Mukherjee, B. N., “Predicting Survival in Burned Patients”, Burns, Vol. 18, No. 5, 1992, pp. 368-372.
12Bull, J. P. and Squire, J. R., ”A Study of Mortality in a Burns Unit.”, Ann. Surg., 1949, pp. 130-160.
13Campbell, M. K., Donner, A., and Webster, K. M., “Are Ordinal Models Useful for Classification?”, Statistics in Medicine, Vol. 10, 1991, pp. 383-394.
14Cooper, D. R. and Emory, C. W., ”Business Research Method”, Chicago, Illinois, Irwin Inc. 1995.
15Coste, J., Walter, E., Wasserman, D., and Venot, A., “Optimal Discriminant Analysis for Ordinal Responses”, Statistics in medicine, Vol. 16, 1997, pp. 561-569.
16Dietterich, T. G., “Approximate Statistical Tests for Comparing Supervised Classification Learning Algorithms”, Neural Computation, Vol. 10, 1998, pp. 1895-1923.
17Dimitras, A.I., Slowinski, R., Susmaga, R. and Zopounidis, C., “Business Failure Prediction using Rough Sets”, European Journal of Operational Research, Vol. 114, 1999, pp. 263-280.
18Ho, W.S., Ying, S.Y. and Burd, A., “Outcome Analysis of 286 Severely Burned Patients: Retrospective Study.”, Hong Kong Med J, Vol. 8, No. 4, August, 2002, pp. 235-239.
19Jie, X., Qingyan, M. and Zheng, W. Y., “Comparable Results between Standardization Methods and Regression Analysis in Predicting Mortality Rate among Samples with Burns”, Burns, Vol. 29, 2003, pp. 247-255.
20Johnson, R. A. and Wichern, D. W., “Applied Multivariate Statistical Analysis”, Prentice-Hall, Upper Saddle River, New Jersey (1998).
21Komorowski, J., Pawlak, Z., Polkowski, L., and Skowron, A., “Rough Sets: Tutorial”, http://www.folli.uva.nl/CD/1999/library/pdf/skowron.pdf.
22Pawlak, Z., “Rough Sets and Intelligent Data Analysis”, Information Sciences, Vol. 147, 2002, pp. 1-12.
23Pawlak, Z., “Rough Sets: Theoretical Aspects of Reasoning about Data”, Kluwer Academic Publishers, 1991.
24Pawlak, Z., Grzymala-Busse, J., Slowinski, R., Ziarko, W., “Rough Sets”, Communications of the ACM, Vol. 38, 1995, pp. 89-95.
25Ryan, C. M., Schoenfeld, D. A., Thorpe, W. P., Sheridan, R. L., Cassem, E. H., and Tompkins, R. G., “Objective Estimates of the Probability of Death from Burn Injuries”, The New England Journal of Medicine. Vol. 338, No. 6, 1998, pp. 362-366.
26Slowinski, R., Zopounidis, C., Dimitras, A.I., “Prediction of Company Acquisition in Greece by Means of the Rough Set Approach”, European Journal of Operation Research, Vol. 100, 1997, pp. 1-15.
27Stevens, James, “Applied Multivariate Statistics for the Social Sciences”, Mahwah, N.J. Lawrence Erlbaum Associates, Inc., 2002.
28Vico, P. and Papillon, J., “Factors Involved in Burn Mortality: a Multivariate Statistical Approach based on Discriminant Analysis”, Burns, Vol. 18, No. 3, 1992, pp. 212-215.
29Walczak, B. and Massart, D. L., “Tutorial Rough Sets Theory”, Chemometrics and Intelligent Laboratory Systems, Vol. 47, 1999, pp.1-16.
30Warsaw University, “RSES 2.2 User’s Guide”, 2005. http://logic.mimuw.edu.pl/~rses
31Yarnold, P. R. and Soltysik, R. C., “Refining Two-Group Multivariable Classification Models Using Univariate Optimal Discriminant Analysis”, Decision Sciences, Vol. 22, 1991, pp. 1158-1164.
32Zöch, G., Schemper, M., Kyral, E., and Meissl, G., “Comparison of Prognostic Indices for Burns and Assessment of their Accuracy”, Burns, Vol. 18, No. 2, 1992, pp. 109-112.