Capturing the local environment and its relationship with health outcomes via geo-localized information extracted from satellite and map services
The local environment surrounding an individual's household is an important predictor of socio-economic status and adverse health events. An increasing number of geo-localized information is being collected, e.g. by Google map services. It is, however, unclear how such a large amount of information can be leveraged to explore the effect of the environment on health. Previous studies have shown that information extracted from Google street view can provide an estimate of neighborhood socio-economic level.
The goal of this project is to extract from satellite and map services novel environmental features and test their association with health outcomes above and beyond simpler measures of environmental exposure and socio-economic status. To extract novel local environment features we will use deep learning methods applied to satellite and street view images as well as more traditional statistical models. This 5-year project will provide a better understanding of how the local environment can impact an individual's health and can inform preventive strategies aimed to reduce environmental risk factors.