We characterize short-term and long-term user engagement patterns in a self-tracking, mobile health app. We introduce and define engagement metrics to capture the quantity, duration, and density of participant engagement according to different domains of self-tracking. We focus our study on Phendo, a research app designed for participants to self-track their experiences of endometriosis ---a chronic disease in women of reproductive age.
Given a cohort of Phendo participants with at least one initial week of use, we analyze their engagement patterns based on amount and timing of daily app usage and stratify them according to their short-term engagement patterns, i.e., within their first 12 weeks of app-use. Given the overall cohort and its stratified short-term participant engagement groups, we then assess overall longitudinal engagement patterns with the app beyond the first 12 weeks, as well as with disease-specific self-tracking domain types.
We identify four groups of participants in the Phendo cohort (n=4,993) according to their short-term engagement: Regulars, Usuals, Occasionals, and Seldoms. Participants across the groups do not differ in disease status or demographics, except for age and education. We find that, while the stratification is based on a short-term period, the groups continue to exhibit distinct longitudinal engagement patterns in the long-term (up to 4 years). We also find that Regulars are more likely to engage with self-tracking related to disease self-management than the other groups.
These findings have implications for the design of research mobile apps, as certain functionalities like tracking of self-management, might yield longer and richer engagement with certain types of participants, even if self-management is not part of the original intent of the research app. More broadly, our proposed engagement metrics and analyses provide a roadmap for exploring participant engagement patterns in mobile research apps.