Application of Poisson Regression Model and Negative Binomial Model on Tuberculosis-A case study of Taichung City

• Main Authors: Chiu, Meng-Chieh, 邱孟婕
• Other Authors: Tong, Lee-Ing
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/9d3663

碩士 === 國立交通大學 === 工業工程與管理系所 === 107 === Global air pollution is becoming more and more serious, causing not only human respiratory related diseases, but also cardiovascular, kidney, lung, and even mental illness. In 2015, the research team at National Taiwan University found that tuberculosis and air pollution are positively related. That is, exposure to PM2.5 of 10μg / m3, the risk of tuberculosis incidence will increase by 39%. This confirms that the air pollution does increase the risk of tuberculosis infection. This study takes the example of Taichung City, which has recently caused serious air pollution due to thermal power plants, to explore outdoor the correlation between outdoor air pollution and tuberculosis. The Poisson and negative binomial regression models are constructed, respectively based on age and gender, the correlation between air pollution and the number of tuberculosis infections is analyzed and the key meteorological and air pollution factors affecting the number of tuberculosis infections are determined. The results of this study found that the number of tuberculosis infections in women was not significantly related to air pollution factors, and the number of tuberculosis infections in men was positively correlated with SO_2 and CO. Assuming that the temperature was fixed, increasing one unit of CO, the number of young men tuberculosis infections cases will increase by 1.96. An increase of one unit of SO_2, the number of old men tuberculosis infections cases will increase by 1.11, which proved that air pollution, meteorological factors and the tuberculosis infections cases of male have a certain degree of correlation. This result is similar to that of a study conducted in South Korea. The reason for the above result is because the airway anatomy, physiological differences between genders, gender hormones, etc., will cause different amounts of inhaled air pollutants. In addition, the regression models established in this study can be utilized to estimate the number of tuberculosis infections per week through air pollution factors and meteorological factors.