|
Abstract
|
SAVIR 2022 Conference Abstracts
Google Street View imagery is a novel and underutilized source of built environment data in violence and injury research. Such data can characterize physical conditions, such as presence of sidewalks, driver visibility, and traffic calming in proximity to motor vehicle collisions and active transportation injuries, or broken sidewalks in relation to falls among older adults. It can also be used to characterize reflections and possible determinants of social conditions, such as sustained disinvestment or physical disorder. Tools to assess such built environment conditions have been developed but remain underused in the injury and violence field, in part due to lack of familiarity among researchers. In this workshop, we will discuss conceptual, legal/ethical, and practical concerns regarding the use of these data. Conceptual issues covered will include theoretical framing of visually assessable features as an influence on behaviors and outcomes, selection of sites to audit and the pros and cons of spatial interpolation, proper consideration of temporality, and the role of historical inequities in developing the conditions being assessed. Legal/ethical issues covered will include the use of the Google API, concerns regarding terms of service and offline caching of images for use with Machine Learning algorithms, and when it is necessary to obtain informed consent from research staff conducting audits. Practical issues will include development of new audit protocols, training of auditors, and availability and quality of imagery. The short course, which has not previously been offered at SAVIR, will include hands-on experience using the Computer Assisted Visual Neighborhood Assessment System (CANVAS), a web-based tool designed to conduct virtual audits of Street View imagery reliably and the de facto standard for conducting such audits for epidemiologic research.
Language: en |