David C. Mohr*, Northwestern University, Chicago, United States Mark Begale, Northwestern University, Chicago, United States Jenna Duffecy, Northwestern University, Chicago, United States Track: Research Presentation Topic: Mobile & Tablet Health Applications Presentation Type: Oral presentation Submission Type: Single Presentation Building: Joseph B. Martin Conference Center at Harvard Medical School Room: Auditorium Date: 2012-09-15 04:45 PM – 05:30 PM Last modified: 2012-09-12

If you are the presenter of this abstract (or if you cite this abstract in a talk or on a poster), please show the QR code in your slide or poster (QR code contains this URL).

Background and Objectives: The aim of our Mobilyze program of research is to develop a context sensing application that can identify in real-time user states relevant to depression and treatment of depression. A growing number of researchers are investigating the use of external, wearable sensors to provide information about user states and activities. However, today’s smartphones come with a rich set of embedded sensors, such as an accelerometer, digital compass, gyroscope, GPS, light sensors, microphone, and camera. These data can be combined with other externally available data such as GIS. Together, data from embedded sensors and external sources can potentially be harnessed to identify user states without requiring users to put on external sensors.
Methods: We will describe the context sensing system, including 1) the low level sensing module that gathers raw sensor data, 2) the feature processing module that processes raw sensor data into usable features, and 3) the classification module that identifies specific user states. The classification module uses supervised machine learning methods to develop individual user algorithms to identify 4 user states: location, activity, social context and mood. We will also describe the Mobilyze mHealth intervention, which is based on behavioral activation. This intervention includes didactic information, tools that support implementation of behavioral activation skills, and feedback graphs such as identified mood by location, activity or social context. The context sensing system can provide real-time outreach to users when it identifies increased risk of worsening mood, or adherence or non-adherence to behavioral activation treatment activities.
Results: We will present the results of our first 8-week field trial using 8 adults with major depressive disorder. Participants showed significant reductions in depression (p<0.001). User satisfaction was high. Promising accuracy rates (60% to 91%) were achieved by learners predicting categorical contextual states (e.g., location). For states rated on scales (e.g., mood), predictive capability was poor. The Mobilyze system is currently being subjected to a major retooling, usability testing, and field testing. Results from this second round of evaluation will be presented.
Conclusions: Mobilyze! is a usable, scalable intervention with preliminary evidence of efficacy.