18.城镇化的低碳、生态与健康效应
Using machine learning to estimate the nonlinear effect of
environmental attributes around residence and school on asthma
occurrence among 35,838 and 12,666 school-age children and
adolescents in Guangzhou and Shenzhen, China
陈奕毅
中山大学
摘要:A growing body of literature indicates that environmental characteristics
surrounding home addresses may have a significant influence on the prevalence of
asthma among children and adolescents. However, the effect of environmental factors
around school on their asthma occurrence is unknown, and our knowledge of the
relative importance and threshold effects of different environmental factors is limited.
Employing data from Guangzhou and Shenzhen collected in 2016, this study utilizes
Gradient Boosting Decision Trees (GBDT) and Partial Dependence Plots (PDP) to
address these gaps. We find potential attenuating effects of relative humidity around
homes and schools on reducing the incidence of asthma, where 74%-78% relative
humidity is the turning point in reducing the risk of asthma. Findings of this study
provide empirical evidence for the importance of improving built and natural
environments around homes and schools in curbing asthma among school-aged
children and adolescents.
关键词:Environmental attributes, gradient boosting decision trees (GBDT), nonlinear
effect, asthma, school-age children and adolescents, China