LAUSR

Abstract Dengue fever (DF) is among the most serious vector-borne diseases and is widespread in tropical and subtropical areas. Ecological environment and socioeconomic drivers have a significant effect on the rapid spread of DF. This study aimed to investigate the long-term relationship between the ecological environment and socioeconomic drivers with DF transmission in the dengue-pandemic city of Guangzhou, China over the period 1998–2016. Principle components analysis was conducted to select key ecological environment and socioeconomic drivers conducive to DF transmission. Statistical models and machine learning regression algorithms, including ordinary least squares analysis, generalized linear and generalized additive models, and support vector regression were utilized to model the occurrence of DF over a long period. Population density, night-time light, travel and land use were selected as key factors. According to the overall performance of the four models, a DF model based on a generalized additive model was chosen as having the best performance. This model not only detected significantly non-linear relationships between key ecological environment and socioeconomic drivers and the occurrence of DF with high correlation coefficient, but also perfectly fitted extreme outbreaks of DF. We have made suggestions about policies and measures regarding its prevention in Guangzhou and other regions where it is endemic from the perspectives of ecological environment and socioeconomic factors. This study aims to shed light on the role of the ecological environment and socioeconomic factors in the transmission of DF and provide the scientific basis and guidance for its future prevention and control.