隨著氣候變遷及全球暖化，使登革熱的疫情有升高的趨勢，登革熱也成為全球最重要的蚊媒傳染疾病，由於典型的登革熱症狀極輕微，類似一般感冒症狀，因此許多人往往不知道自己感染登革熱而忽視此疾病，但是當登革熱患者第二次感染登革熱時，身體會產生劇烈的免疫反應，使病患有嚴重的症狀且極為痛苦，死亡率更高達20%。因此，登革熱患者是否造成不良預後是一個很重要的議題。
2015年登革熱在臺南大爆發，造成民眾十分恐慌，為此，如何事先預防以及罹患登革熱後如何照護是當今重要的醫學議題之一，因此，本研究目的以臺南市2015年登革熱確診個案通報狀況為例，透過分析登革熱病患之人口學特性以及就醫習慣，臨床症狀、發病日與通報日時間差，分析有無統計上差異，並探討初期臨床症狀因素對登革熱患者造成不良預後是否相關，以貝氏分類法預測登革熱不良預後之高危險群，並透過商業智慧相關工具將資料處理以視覺化的方式呈現。
本研究結果顯示運用貝氏分類法預測正確率有95%以上，分析結果老年人較容易有不良預後的情況，有噁心嘔吐、腸胃道症狀、倦怠、白血球減少、頭暈、血小板低下、出血、消化道出血、呼吸窘迫、胸悶胸痛、休克、肝功能異常、泌尿系統損傷、意識改變、心律異常或嚴重出血症狀有不良預後比率比沒有不良預後比率高，可以作為醫護人員臨床診斷是否為不良預後之參考。

SUMMARY
In the 2015 dengue outbreak in Tainan, causing panic in people. We do not know which type of dengue patients are prone to have adverse prognosis. Therefore, the aim of this study is to analyze the status of dengue fever diagnosed in 2015 in Tainan city. By analyzing dengue fever demographic characteristics, medical habits, clinical symptoms, the time difference between the date of onset and the notification time to test the statistical differences among them. Then, to explore whether the initial clinical symptoms and signs of dengue fever is related to patients with adverse prognosis. In addition, the Bayesian classification is used to predict the high risk group of adverse prognosis of dengue. Finally, we visualize the data to reveal the implied insights by business intelligence tools. The results of this study showed that the correct rate of prediction is more than 95% by Bayesian classification. Furthermore, old people are more likely to have adverse prognosis such as nausea, vomiting, gastrointestinal symptoms, fatigue, leukopenia, dizziness, thrombocytopenia, bleeding, gastrointestinal bleeding, respiratory distress, chest tightness, abnormal liver function, etc. The conclusion of this study indicated that the majority of patients with adverse prognosis were elderly patients whose clinical symptoms were less pronounced and the disease progressed rapidly during the dengue outbreak in 2015 in Tainan.
INTRODUCTION
Since 1987, the dengue fever epidemic has become the most important mosquito-borne disease in Taiwan. In 2014, the number of local cases in Taiwan increased dramatically to 15,464 cases (reported by day), and the number of domestic cases in Taiwan in 2015 was 43,348 cases (reported by day), which is the highest over the past. In Tainan city, there were only 155 local cases in 2014, however, it experienced the most serious dengue fever in 2015 with 22,777 indigenous cases.

The aim of this study is to analyze the status of dengue fever diagnosed in 2015 in Tainan city. We tested the statistical differences among demographic characteristics of dengue patients such as gender, age, type of living area, medical treatment, clinical symptoms, time between onset and notification. Then, to explore whether the initial clinical symptoms and signs of dengue fever is related to patients with adverse prognosis. The results are expected to provide references for the medical staff in Tainan area, and help them early detection and treatment of dengue fever patients and
reduce the mortality rate.

MATERIALS AND METHODS
The data source of this study is according to the information of dengue fever in 2015 in Tainan city. There were 22,777 cases, of whom 337 admitted to ICU, 189 deaths and 298 had adverse prognosis. We extracted some factors such as gender, age, status at notification, history of chronic diseases, symptoms, time between onset and notification and analyzed the relationships among them. The adverse prognosis refers to patients infected with dengue fever admitted to ICU or death, excluding notification has been admitted to ICU or death.

Microsoft Access was used to clean and integrate the data, initially, we analyzed the data and constructed the data model by Microsoft Excel to accelerate the data processing, analysis and query. Then a business intelligence software, Tableau was applied to combine and visualize the data. Furthermore, this study used SPSS, a kind of statistical software to implement data analysis: (1) descriptive statistics, (2) chi-square test, (3) regression analysis.

In this study, Bayesian classification was used to predict the group with high risk of dengue adverse prognosis. The causal relationship between clinical symptoms and adverse prognosis of dengue was explored to provide clinical judgment references for medical staff. In addition, showed the status of dengue notification data through online analytical processing (OLAP), which can analyze the data from different dimensions and facilitate to detect the high risk group with adverse prognosis of dengue fever.

RESULTS AND DISSCUSSION
The results of this study showed that using the Bayesian classification method to predict the correct rate of more than 95%,

The results showed that the correct rate of prediction is more than 95% by Bayesian classification method. Also, the elderly is more likely to have adverse prognosis such as nausea, vomiting, gastrointestinal symptoms, fatigue, leukopenia, dizziness, thrombocytopenia, bleeding, gastrointestinal bleeding, respiratory distress, chest tightness, pain, shock, abnormal liver function, urinary tract injury, altered consciousness, accumulation of body fluids or severe bleeding symptoms. It means that the rate of old people with adverse prognosis are more than those with no adverse prognosis. Therefore, it can be taken as reference of clinical diagnosis for healthcare workers in the future.
 
CONCLUSION
The conclusion of this study was that older people and patients with severe and non-specific initial symptoms at time of reporting had higher risk of ICU admission and mortality during the dengue outbreak in 2015 in Tainan. The majority of patients with adverse prognosis were elders whose clinical symptoms were less pronounced and the disease progressed rapidly. The findings of this study can be provided to first-line healthcare staffs to early identify patients who may be admitted to ICU or be fatal. In the future, we should pay more attention to the early identification of early atypical symptoms, especially for the elderly, in order to improve the quality of care in patients with dengue.

中文文獻
尹相志(2009)，SQL Server 2008 OLAP多維度分析，台北市：悅知文化。
胡坤德(1995)，貝氏可能值推測方法之研究，東海學報，第三十六卷第三期，頁95-116。
黃旭愉(2005)，決策分析的利器—線上分析處理(OLAP)，華南銀行資訊室。
黃高彬、韓明榮、劉永慶、劉清泉、劉建衛、張肇松、蔡季君、曾逸豪、蔡森郎(2002)，登革熱/登革出血熱臨床症狀、診斷與治療，台北市：行政院衛生署疾病管制局。
英文文獻
Biere, M. (2003). Business Intelligence for the Enterprise: Prentice Hall Professional.
Blanco, R., Inza, I., Merino, M., Quiroga, J., & Larrañaga, P. (2005). Feature selection in Bayesian classifiers for the prognosis of survival of cirrhotic patients treated with TIPS. Journal of Biomedical Informatics, 38(5), 376-388.
Bresfelean, V. P., Bresfelean, M., Ghisoiu, N., & Comes, C. A. (2007). Data Mining Clustering Techniques in Academia. Paper presented at the the 9th International Conference on Enterprise Information Systems, Funchal, Madeira - Portugal.
Centers for Disease Control, M. o. H. a. W., R.O.C. (Taiwan). (2016). Guidelines for Dengue / Chikungunya / Zika Virus Infection Control. Taipei City, Taiwan: Centers for Disease Control, Ministry of Health and Welfare, R.O.C. (Taiwan).
Cestnik, B., Kononenko, I., & Bratko, I. (1987). Assistant 86: A knowledge-elicitation tool for sophisticated users. Paper presented at the Proceedings of the 2nd European Conference on European Working Session on Learning, Yugoslavia, Europe.
Chen, C. M., Chan, K. S., Yu, W. L., Cheng, K. C., Chao, H. C., Yeh, C. Y., & Lai, C. C. (2016). The outcomes of patients with severe dengue admitted to intensive care units. Medicine, 95(31), e4376.
Chiang, T. A., & Che, Z. H. (2010). A fuzzy robust evaluation model for selecting and ranking NPD projects using Bayesian belief network and weight-restricted DEA. Expert Systems with Applications, 37(11), 7408-7418.
Codd, E. F., Codd, S. B., & Salley, C. T. (1993). Providing OLAP (on-line analytical processing) to user-analysts: An IT mandate. Codd and Date, 32.
Cui, G., Wong, M. L., & Lui, H. K. (2006). Machine learning for direct marketing response models: Bayesian networks with evolutionary programming. Management Science, 52(4), 597-612.
Davenport, T. H. (2013). Analytics 3.0. Harvard Business Review, 91(12), 64-72.
Deas, T. S., Bennett, C. J., Jones, S. A., Tilgner, M., Ren, P., Behr, M. J., Stein, D. A., Iversen, P. L., Kramer, L. D., & Bernard, K. A. (2007). In vitro resistance selection and in vivo efficacy of morpholino oligomers against West Nile virus. Antimicrobial Agents and Chemotherapy, 51(7), 2470-2482.
Duda, R. O., & Hart, P. E. (1973). Pattern Classification and Scene Snalysis. New York: Wiley.
Ferreira, L., & Borenstein, D. (2012). A fuzzy-Bayesian model for supplier selection. Expert Systems with Applications, 39(9), 7834-7844.
Gubler, D. J. (1998). Dengue and dengue hemorrhagic fever. Clinical Microbiology Reviews, 11(3), 480-496.
Gubler, D. J. (2002). Epidemic dengue/dengue hemorrhagic fever as a public health, social and economic problem in the 21st century. Trends in Microbiology, 10, 100-103.
Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., & Witten, I. H. (2009). The WEKA data mining software: an update. ACM SIGKDD Explorations Newsletter, 11(1), 10-18.
Han, J. W., Pei, J., & Kamber, M. (2011). Data Mining: Concepts and Techniques. Burlington, Massachusetts, United States: Morgan Kaufmann Publishers.
Hsieh, C. C., Cia, C. T., Lee, J. C., Sung, J. M., Lee, N. Y., Chen, P. L., Kuo, T. H., Chao, J. Y., & Ko, W. C. (2017). A cohort study of adult patients with severe dengue in Taiwanese intensive care units: the elderly and APTT prolongation matter for prognosis. PLOS Neglected Tropical Diseases, 11(1), e0005270.
Hsu, J. C., Hsieh, C. L., & Lu, C. Y. (2017). Trend and geographic analysis of the prevalence of dengue in Taiwan, 2010–2015. International Journal of Infectious Diseases, 54, 43-49.
Karyanti, M. R., Uiterwaal, C. S., Kusriastuti, R., Hadinegoro, S. R., Rovers, M. M., Heesterbeek, H., Hoes, A. W., & Bruijning-Verhagen, P. (2014). The changing incidence of dengue haemorrhagic fever in Indonesia: a 45-year registry-based analysis. BMC Infectious Diseases, 14(1), 412.
Liew, S. M., Khoo, E. M., Ho, B. K., Lee, Y. K., Omar, M., Ayadurai, V., Yusoff, F. M., Suli, Z., Mudin, R. N., & Goh, P. P. (2016). Dengue in Malaysia: factors associated with dengue mortality from a national registry. PLOS One, 11(6), e0157631.
Low, J. G., Ong, A., Tan, L. K., Chaterji, S., Chow, A., Lim, W. Y., Lee, K. W., Chua, R., Chua, C. R., & Tan, S. W. (2011). The early clinical features of dengue in adults: challenges for early clinical diagnosis. PLOS Neglected Tropical Diseases, 5(5), e1191.
Limkittikul, K., Brett, J., & L'Azou, M. (2014). Epidemiological trends of dengue disease in Thailand (2000–2011): a systematic literature review. PLOS Neglected Tropical Diseases, 8(11), e3241.
Mitchell, T. M. (1997). Machine Learning. New York: McGraw-Hill.
Moraes, G. H., de Fátima Duarte, E., & Duarte, E. C. (2013). Determinants of mortality from severe dengue in Brazil: a population-based case-control study. The American Journal of Tropical Medicine and Hygiene, 88(4), 670-676.
Ong, A., Sandar, M., Chen, M. I., & Sin, L. Y. (2007). Fatal dengue hemorrhagic fever in adults during a dengue epidemic in Singapore. International Journal of Infectious Diseases, 11(3), 263-267.
Palaniappan, S., & Ling, C. (2008). Clinical decision support using OLAP with data mining. International Journal of Computer Science and Network Security, 8(9), 290-296.
Pang, J., Thein, T.-L., Leo, Y.-S., & Lye, D. C. (2014). Early clinical and laboratory risk factors of intensive care unit requirement during 2004–2008 dengue epidemics in Singapore: a matched case–control study. BMC Infectious Diseases, 14(1), 649.
Perina, A., Cristani, M., Xumerle, L., Murino, V., Pignatti, P. F., & Malerba, G. (2009). Fully non-homogeneous hidden Markov model double net: A generative model for haplotype reconstruction and block discovery. Artificial Intelligence in Medicine, 45(2), 135-150.
Pinto, R. C., de Castro, D. B., de Albuquerque, B. C., de Souza Sampaio, V., dos Passos, R. A., da Costa, C. F., Sadahiro, M., & Braga, J. U. (2016). Mortality predictors in patients with severe dengue in the State of Amazonas, Brazil. PLOS One, 11(8), e0161884.
Pourabbas, E., & Rafanelli, M. (2000). Hierarchies and relative operators in the OLAP environment. ACM Sigmod Record, 29(1), 32-37.
Sell, D., Cabral, L., Motta, E., Domingue, J., & Pacheco, R. (2005). Adding semantics to business intelligence. Paper presented at the Database and Expert Systems Applications: 16th International Conference, Copenhagen, Denmark.
Simmons, C. P., Farrar, J. J., van Vinh Chau, N., & Wills, B. (2012). Dengue. New England Journal of Medicine, 366(15), 1423-1432.
Teixeira, M. G., & Barreto, M. L. (2009). Diagnosis and management of dengue. BMJ, 339, b4338.
Thanh, L. V., Choisy, M., Bryant, J. E., Duoc, V. T., Thai, P. Q., Horby, P., Hien, N. T., Kieu, H., Trung, N. V., & Kinh, N. V. (2015). A dengue outbreak on a floating village at Cat Ba Island in Vietnam. BMC Public Health, 15(1), 940.
Thein, T. L., Leo, Y. S., Fisher, D. A., Low, J. G., Oh, H. M., Gan, V. C., Wong, J. G., & Lye, D. C. (2013). Risk factors for fatality among confirmed adult dengue inpatients in Singapore: a matched case-control study. PLOS One, 8(11), e81060.
Thomas Jr, J. H. (2001). Business intelligence–why. eAI Journal, 2001, 47-49.
Vitt, E., Misner, S., & Luckevich, M. (2002). Business Intelligence: Making Better Decisions Faster: Microsoft Press.
Wang, S. F., Chang, K., Loh, E. W., Wang, W. H., Tseng, S. P., Lu, P. L., Chen, Y. H., & Chen, Y. M. (2016). Consecutive large dengue outbreaks in Taiwan in 2014–2015. Emerging Microbes & Infections, 5(12), e123.
WHO. (2009). Dengue: Guidelines for Diagnosis, Treatment, Prevention and Control Chapter 1: Epidemiology, burden of disease and transmission. Geneva: Stylus Pub Llc.