JMIR Publications

ABSTRACT

Background:

Mobile health technology demonstrated the ability of smartphone apps and sensors to collect data pertaining to patient activity, behavior, symptoms, cognition, and context. Mobile sensing, in particular, has the ability to collect data objectively and continuously during the lived experience of individuals. We examined impulsivity because it is an underlying factor in many today’s major public health concerns (e.g., obesity) and because it is a promising and challenging context in which to examine the value of mobile sensing.

Objective:

Explore the feasibility of using mobile sensing data to assess and monitor impulsivity passively. The goal of this study was to understand relationships among self-report measures of impulsivity, and mobile phone-based objective passive measures assessed via a cross platform mobile sensing application.

Methods:

As part of the Digital Marshmallow Test (DMT) study, we enrolled 26 participants to a study app (mPulse) with mobile sensing capabilities over 21 days on both Android and iOS platforms. The mobile sensing system collected data from call logs, battery charging and screen status. To validate the model, we used mobile sensing features to predict common clinical self-reports, objective behavioral measures, and ecologically momentary assessments of impulsivity and related contacts (such as risk taking, attention, and affect).

Results:

Overall, findings suggest that passive measures of mobile phone use such as battery usage, screen status, and call logs can predict different facets of trait and state impulsivity. For trait impulsivity, the models significantly explained variance in sensation, planning, and perseverance traits but failed to explain motor, urgency, premeditation, and attention traits. Passive sensing features from call logs, battery charging, and screen status were particularly useful in explaining and predicting trait-based sensation seeking. On the daily level, the model successfully predicted objective behavioral measures such as present bias in delayed discounting tasks, commission and omission errors in a cognitive attention task, and total gains in a risk-taking task. Our models also predicted daily ecological momentary assessment (EMA) questions on positivity, stress, productivity, healthiness, and emotion/affect. Perhaps most intriguingly, the model failed to predict daily EMA designed to measure previous-day impulsivity using face valid questions.

Conclusions:

The study demonstrated the potential for developing trait and state impulsivity phenotypes from everyday mobile phone sensors. Suggestions for building more precise passive sensing models and limitations of the current research are discussed. Clinical Trial: ClinicalTrials.gov NCT03006653