Grant: Planning/Dissemination Grants - IPPH - ICS - Artificial Intelligence - General

In order to better understand how urban form relates to health, it is important to obtain reliable measure of street characteristics, in particular those promoting healthy behaviors (e.g. active transport) and preventing traffic injuries, such as traffic calming measures. Virtual environmental audits with Google Street View images have proven valuable to document urban characteristics and urban change. Google Street view enables researchers (or crowdsourcing participants) to code various built environment characteristics online, instead of having to conduct direct observations in the streets. Moreover, Google Street View also provides historical images since 2007, enabling the detection of change in urban features over time. Coding the images manually is however very time-consuming and costly. Automatizing the virtual environmental audits has the potential of providing uniform measures over large geographical areas, or even globally. Previous studies have developed artificial intelligence algorithms to automatically detect objects (e.g. traffic signs) in Google Street View images. Using similar types of images, scientists have developed artificial intelligence methods to detect a range of street characteristics for example for self-driving cars or to help people with visual impairments in traffic situations (e.g. indication of street crossings). The aim of this proposal is to build a network of researchers in the fields of Artificial Intelligence, Public Health, and Urban Planning, and to plan a project on the use Google Street View images to detect changes in urban form characteristics. The objectives of the project are to obtain adequate, reliable and cost-efficient measures of urban transformations that can then serve to health researchers interested in analyzing how urban change relates to health behavior, health, and health inequities. The codes used and data will be made available on open-source platforms for researchers to freely use in future projects.