Deep learning for identifying environmental risk factors of acute respiratory diseases in Beijing, China: implications for population with different age and gender

This study focuses on identifying environmental health risk factors related to acute respiratory diseases using deep learning method. Based on respiratory disease data, air pollution data and meteorological environmental data, cross-domain risk factors of acute respiratory diseases were identified in Beijing, China. We conducted age and gender stratified deep neural network models in air pollution epidemiology. We ranked risk factors of respiratory diseases in stratified populations and conducted quantitative comparison. People >= 50 years were more sensitive to PM2.5 pollution than <50 years people, especially women >= 50 years. Compared with women, both men >= 50 years and <50 years were more susceptible to PM10. Young women <50 years were more sensitive to general air pollutants such as SO2 and NO2 than <50 years young men. Meteorological factors such as wind speed and precipitation could promote the diffusion of fine particulate matter and general air pollutants (SO2, NO2, etc.), which could help to reduce the incidence of acute respiratory diseases. This study represents a quantitative analysis of environmental health risk factors identification related to acute respiratory diseases based on deep neural network method. The results of this study could help people to improve their awareness of acute respiratory diseases prevention.