SARA: An App for Engaging Users in Mobile Health Data Collection

I recently read the paper “SARA: A Mobile App to Engage Users in Health Data Collection.” [1]  The problem they address is that in mHealth, one wants to collect data that requires self-report or effort on the part of the user to log.  This has high burden.  For example, in the ecological momentary assessment, participants receive questions about mood one or more times a day.  They also may be asked to report alcohol/marijuana/cigarette use.  One approach to handling this is to find more automated ways to collect this data: for instance, in PuffMarker, they attempt to detect cigarette puffs from a wrist-worn sensor. [2]  Another approach is to increase user engagement.  One way to increase engagement is to pay users more, but this is not scalable.  Ideally one would like methods where the cost does not scale linearly with the number of users.  SARA provides an app with data collection capabilities along with a set of engagement strategies, each based on psychological theory.

SARA stands for “Substance Abuse Research Assistant.”  The app collects data from adolescents and adults from 14-24 who have reported bige drinking and/or marijuana use recently.  Users are asked to complete daily surveys with seven questions about mood, along with two active tasks that test reaction time, problem solving skills, and a few other abilities.  It collects movement and geolocation data in the background.  It also allows for visualization of collected data.

In order to engage users, they use 10 different engagement strategies grounded in theories of motivation and persuasion.  The main criteria for deciding on these engagement strategies are that they are feasible to implement, consistent with the aquarium theme of SARA, and culturally appropriate.

The app is based on the idea of an aquarium.  As more surveys and tasks are completed, the size of the aquarium grows.  Also, results of the various engagement strategies are displayed.

My comments: Engagement in collecting self-report data is a very important issue.  If you’ve ever tried actually filling out surveys, you know how boring they are, and doing this daily or multiple times a day is far from ideal.  I think such an app can provide value in several ways:

  1. As a platform for researchers across various disciplines to collect data about adolescents and young adults.  Clearly some of the questions would need changing, but having a common framework for collecting EMA data for adolescents and young adults would allow the studies to control for the collection method.  It’s quite possible that the interface itself would affect the responses to questions.
  2. Allowing study designers to keep users engaged: reducing dropout among study participants would be important in two ways.  One is that it increases the sample size of the study.  Two is that those participants most at risk of dropout may have a very different substance use trajectory or mood trajectory from other users, and thus having them drop out of the study actually leads to biased population-level summaries of the data.
  3. As a way to test different engagement strategies and describe users.  They mention this, but the possibilities are large.  One could test different temporal recommendation strategies (for instance, contextual bandits vs. some matrix-based recommender techniques).  One could model how much specific types of participants are likely to cost based on static covariates.  They have a follow-up paper that I haven’t looked at yet that does a study with some statistical analysis, so I look forward to reading that in the near future.

[1] Rabbi, Mashfiqui, Meredith Philyaw-Kotov, Jinseok Lee, Anthony Mansour, Laura Dent, Xiaolei Wang, Rebecca Cunningham et al. “SARA: a mobile app to engage users in health data collection.” In Proceedings of the 2017 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2017 ACM International Symposium on Wearable Computers, pp. 781-789. ACM, 2017.

[2] Saleheen, Nazir, Amin Ahsan Ali, Syed Monowar Hossain, Hillol Sarker, Soujanya Chatterjee, Benjamin Marlin, Emre Ertin, Mustafa Al’Absi, and Santosh Kumar. “puffMarker: a multi-sensor approach for pinpointing the timing of first lapse in smoking cessation.” In Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing, pp. 999-1010. ACM, 2015.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.