Text Message Analysis Using Machine Learning to Assess Predictors of Engagement With Mobile Health Chronic Disease Prevention Programs: Content Analysis

Abstract

Background SMS text messages as a form of mobile health are increasingly being used to support individuals with chronic diseases in novel ways that leverage the mobility and capabilities of mobile phones. However, there are knowledge gaps in mobile health, including how to maximize engagement. Objective This study aims to categorize program SMS text messages and participant replies using machine learning (ML) and to examine whether message characteristics are associated with premature program stopping and engagement. Methods We assessed communication logs from SMS text message–based chronic disease prevention studies that encouraged 1-way (SupportMe/ITM) and 2-way (TEXTMEDS [Text Messages to Improve Medication Adherence and Secondary Prevention]) communication. Outgoing messages were manually categorized into 5 message intents (informative, instructional, motivational, supportive, and notification) and replies into 7 groups (stop, thanks, questions, reporting healthy, reporting struggle, general comment, and other). Grid search with 10-fold cross-validation was implemented to identify the best-performing ML models and evaluated using nested cross-validation. Regression models with interaction terms were used to compare the association of message intent with premature program stopping and engagement (replied at least 3 times and did not prematurely stop) in SupportMe/ITM and TEXTMEDS. Results We analyzed 1550 messages and 4071 participant replies. Approximately 5.49% (145/2642) of participants responded with stop, and 11.7% (309/2642) of participants were engaged. Our optimal ML model correctly classified program message intent with 76.6% (95% CI 63.5%-89.8%) and replies with 77.8% (95% CI 74.1%-81.4%) balanced accuracy (average area under the curve was 0.95 and 0.96, respectively). Overall, supportive (odds ratio [OR] 0.53, 95% CI 0.35-0.81) messages were associated with reduced chance of stopping, as were informative messages in SupportMe/ITM (OR 0.35, 95% CI 0.20-0.60) but not in TEXTMEDS (for interaction, P<.001). Notification messages were associated with a higher chance of stopping in SupportMe/ITM (OR 5.76, 95% CI 3.66-9.06) but not TEXTMEDS (for interaction, P=.01). Overall, informative (OR 1.76, 95% CI 1.46-2.12) and instructional (OR 1.47, 95% CI 1.21-1.80) messages were associated with higher engagement but not motivational messages (OR 1.18, 95% CI 0.82-1.70; P=.37). For supportive messages, the association with engagement was opposite with SupportMe/ITM (OR 1.77, 95% CI 1.21-2.58) compared with TEXTMEDS (OR 0.77, 95% CI 0.60-0.98; for interaction, P<.001). Notification messages were associated with reduced engagement in SupportMe/ITM (OR 0.07, 95% CI 0.05-0.10) and TEXTMEDS (OR 0.28, 95% CI 0.20-0.39); however, the strength of the association was greater in SupportMe/ITM (for interaction P<.001). Conclusions ML models enable monitoring and detailed characterization of program messages and participant replies. Outgoing message intent may influence premature program stopping and engagement, although the strength and direction of association appear to vary by program type. Future studies will need to examine whether modifying message characteristics can optimize engagement and whether this leads to behavior change.

Text Messages to Improve Medication Adherence and Secondary Prevention After Acute Coronary Syndrome: The TEXTMEDS Randomized Clinical Trial

“…Separate analyses indicated that supportive and informational messages in particular elicited more message replies. 31 There are several limitations to our study that need to be considered. The study intervention, although evaluated in a randomized design, was not able to be blinded.…”

“…Separate analyses indicated that supportive and informational messages in particular elicited more message replies. 31…”

The Goldilocks Dilemma on Balancing User Response and Reflection in mHealth Interventions: Observational Study

“…Less research has compared specific types of content and has rarely tested different types within the same study. An exception is a recent study by Klimis et al [ 15 ], wherein they used machine learning to demonstrate that text messages with informative (providing health facts or education) and instructional (providing tips or recommendations) message intents were associated with increased engagement, while notification messages that addressed noneducational matters (eg, welcome and exit messages) were associated with reduced engagement [ 15 ]. Our study targeted a two-way message and varied the levels of reflection and response burdens in that message.…”

“…Across the mHealth literature, engagement tends to be highly variable [10,11], which has spurred a whole body of research that aims to understand predictors of engagement, including user characteristics and intervention features (eg, intervention duration and frequency of sending content) [10][11][12][13][14]. However, very little research has attended to the type of mHealth content that users are expected to engage with [15] and, more specifically, how the content may be requesting more or less cognitive reflection. The primary goal in having users engage with mHealth content is health behavior change.…”

Development of a patient‐centered text message‐based platform for the self‐management of interstitial cystitis/bladder pain syndrome symptoms

“…19 The following metrics were defined a priori based on prior studies and analyzed quantitatively using proportions. 20,21 They also served as our primary and secondary outcomes.…”