An Application of Decision Tree on Multiple Sclerosis
| Autor: | Hui-Chuan Lu, 盧惠銓 |
|---|---|
| Rok vydání: | 2013 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 101 Background : Being a kind of rare disease in Taiwan, Multiple Sclerosis (MS) has the highest prevalence of rare diseases. According to the data of the Department of Health, not only does the medical spending rise, but also the number of people diagnosed with MS is on the increase. Early diagnosis can be difficult because of unknown etiology and various onset of symptoms. MS affects both young people and older adults. MS itself is not considered a fatal disease, but it limits one’s ability to exercise or numbness. And that has made a heavy burden in the lives of the MS patients or their families and even the social medical insurance. But there aren’t many domestic researches in MS. MS patients can only take the disease-modifying treatments or rehabilitation treatments. It’s important for patients to obtain early diagnosis and treatment which can help reduce expenditures on medical costs, and reduce stress and feelings of burden.Therefore, identifying the risk factors for MS is a priority work. Objectives: 1. Using the Classification and Regression Tree (CART) of data mining prediction techniques to construct a classification model of risk factors for MS. 2. Discussion on the risk factors associated with MS. Methods: The data sources are from Taiwan’s National Health Insurance Research Database (2003~2006). Based on a case-control study, the analysis works with 1:4 matching conditional on both the area and the degree of urbanization. And identify the case group on a group of 303 MS-sufferers, and the case-control group on a group of 1,212 non-MS-sufferers. These samples based on gender, age, Elixhauser Comorbidities for statistical analysis. Using of Classification and Regression Trees (CART) for decision tree analysis to filter out MS patients’s risk factors. Predict the factors that cause etiology of MS. Results : 1. People older than 66.5 years old have less chances to suffer from MS. (P = 0.026). 2. People who are younger than 66.5 years old but have suffered neurological diseases over the past three years have more chances to suffer from MS (P = 0.023). 3. Males who are younger than 66.5 years old and have never suffered from neurological diseases over the past three years have less chances to suffer from MS (P = 0.012). 4. Females who are younger than 66.5 years old and have suffered from both neurological diseases and peptic ulcer disease excluding bleeding have more chances to suffer from MS (P = 0.010). 5. Females who are younger than 54.5 years old and have suffered from both neurological diseases and peptic ulcer disease excluding bleeding have more chances to suffer from MS than those who are aged between 54.5 to 66.5. (P = 0.010 ). Conclusion : In Taiwan, the patients who first time be diagnosed with MS has more serious medical symptoms. In these cases, they are thought that there’s a combination of many factors like almost uncountable and various symptoms of MS as well as the differences between domestic patterns and foreign ones. The decision model established in this study presented good predictive capability at both ROC Curve and Odds Ratio. This decision model can provide a basis for judging the clinical diagnosis and allow physicians to diagnose accurately and give the patients treatments. And with the appropriate treatment, it can help stop or slow the rate of further deterioration of the illness, reduce health care costs, and extend the survival time of patients. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|