Using Wearable Cameras in Your Research
Wearable cameras automatically capture images from a first-person point of view. Examples include the Microsoft SenseCam, Vicon Autographer, and Google Glass. The March 2013 theme issue on “Wearable cameras in health” in the American Journal of Preventive Medicine highlighted that wearable cameras can enhance aspects of our understanding of lifestyle behaviors such as nutrition, sedentary behaviour, and active travel, which are related to health and disease outcomes. The collection of six articles in this theme issue may well motivate new studies using wearable cameras to better understand the type and context of individual and population lifestyle behaviors. The aim for this blog post is to give researchers a “cheat sheet” on using wearable cameras.
Constructing a study aim: Researchers may find it helpful to explore past research on the SenseCam wiki, the Microsoft SenseCam research webpage, or through some prolific authors in the area such as Hodges, Berry, Smeaton, Foster, Moulin, Gurrin, Kelly, and others.
Ethics clearance: Kelly’s article provides an ethical framework for conducting wearable camera health behavior research with informed consent of participants being critical. There has been precedence of using wearable cameras in children (Cowburn, Kelly), adults (Kelly, Doherty), and older adults (Caprani)
Data collection: A freely available open-source SenseCam browser should be used to initialize Microsoft and Vicon devices following technical document guidelines. Participants should be given an information leaflet and also a briefing in case they are asked ethical/privacy questions by others they encounter (Kelly). In some circumstances participants can be compensated for their time (Kerr).
Data analysis: The SenseCam browser is the best tool to analyze wearable camera data to identify the type and context of behaviors participants were engaged in. There is a series of help videos, and also guidance on behavior annotation schemes (Kerr, Doherty). The annotated set of behaviors can be exported to CSV format (see item 26 here), which then allows researchers to analyze the behaviors in their statistical software of preference.
– Aiden R. Doherty, PhD
University of Oxford & Dublin City University
*AD is a Marie Curie postdoctoral research fellow supported by the Irish Health Research Board under grant number MCPD/2010/12.