Forecasting Outpatient Visits for Influenza-like Illness—The Case of Taichung Influenza-like Illness Outpatient Visits
| Autor: | Jui-Ying Tsai, 蔡瑞瑩 |
|---|---|
| Rok vydání: | 2017 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 105 Influenza-like-illness (ILI) is one of the most changeable and wide-spread epidemic. During flu seasons, the variability of outpatient visits could affect medical resources in hospitals seriously. Nonetheless, up to date, most of the ILI-related researches are focused on the entire trend, making it difficult apply to regional demands. Hence, our study choose the city which involves the highest ILI outpatient number in Taiwan, trying to forecast the approaching outpatient visits. We anticipate that our study could benefit hospitals for medical resources preparation (e.g. pharmaceuticals inventory management and human resources assignment). For the sake of improving forecast accuracy, our study proposes forecast models by adding various aspects of parameters, including time-series factor, geographical factor, weather factor, air pollution factor and the length of historical data. After analyzing the correlation between above factors and ILI outpatient visits of our target city, we forecast the approaching weekly outpatient visits in terms of the result, comparing forecast accuracy and then determine the best forecast model. We propose four kinds of forecast models, including time-series model, geographical correlation model, multiple regression model and mixed model. The sources of data in our study comes from two terms of real-time open-data of authority; that is to said, the forecast value of our study could be updated with official data. We figure out that the forecast accuracy could be improved by adding geographical factor; moreover, our study concludes that the mixed model which involves geographical factor, air pollution factor and specific length of historical data could gain the best accuracy competence. Afterwards, we discuss about the result and conclude that the weak effect of weather factor is due to the less climate variability in Taiwan. To sum up, according to the standard of forecast index proposed from the scholar, our model forecasts well, which may be able to be taken into account by target city. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|