I just wanted to post a quick note letting folks know that I’ll be presenting at a symposium at the 8th Scientific Meeting for the International Society for Research on Internet Interventions (ISRII) from April 7 to 9 in Seattle. My presentation is titled “Scalable Data Collection in the Wild: Clinical and Practical Challenges” and will discuss some of the challenges researchers face when doing smartphone data collection in the wild. (Sound familiar?)

In addition to giving the presentation, I'm also targeting this event to release the first public version of PassiveDataKit so that in addition to identifying and describing challenging aspects of mobile data collection, I have a solution in-hand to address those challenges. This commits me to a somewhat ambitious development schedule, but one that's eminently doable.

Abstract

Leveraging smartphone and mobile device sensors to create a more comprehensive understanding of end-users has become a popular activity in modern research studies and interventions. However, constantly shifting technology foundations, evolving end-user expectations, and changes in popular sentiment feed the endless challenge of successfully collecting passive data in the wild. The traditional approach to dealing with these issues has been to create as fixed an environment as possible, artificially minimizing changes to software and limiting subjects to pre-selected devices in order to maintain a stable environment during lengthy clinical trials. Unfortunately, this approach often ignores the subjects’ device preferences – adversely affecting data collection when information is not collected on the user’s primary device – and changes in technology, which raises questions about the replicability and relevance of findings from recently-obsoleted study platforms.

We present some of our successes, failures, and lessons learned when shifting to a product-based perspective that focused on meeting the engineering challenges of keeping up with the rapidly evolving mobile technology ecosystem. We demonstrate the necessity of keeping engineers engaged throughout the research process, not only during the technology implementation phase. We describe the design and rationale for technical support infrastructure that is not involved in the primary data collection, but provides a basis for identifying and addressing technology issues before the primary data mission can be irrevocably compromised. Protecting sensitive data from subjects is a key requirement for this genre of research and we compare different privacy-preservation approaches as applied to competing research interests. Finally, we discuss how clinical projects must embrace continuous iteration throughout a research study, and how this increased alignment with real-world technology ecosystems and processes improves existing studies and creates new opportunities for more impactful research.

Submission Coauthors

Sohrob Saeb, Kathyn Noth Tomasino, Emily Lattie, Marya Corden, Mark Begale