A critical problem in mental healthcare is determining whether and in what ways a patient has changed as a result of the intervention. In addition, the ability to predict whether a patient will complete and benefit from treatment helps practitioners customize treatment prescriptions more effectively and administer timely treatment modifications. Existing research has examined demographic and medical background predictors but has not explored variables reflecting daily behavior patterns and behavioral context for treatment module completion (e.g., variation in location and time when completing treatment modules).
In this project, we propose that smartphones can be used to obtain such evidence, as smartphones have shown great utility in monitoring key aspects of daily behavior such as mobility, physical activity, social interaction, and device usage. We aim to determine whether we can, more accurately than using baseline predictors identified by previous studies, predict which patients adhere to and benefit from treatment more than others. We plan to (1) characterize the smartphone sensed behavioral patterns, how they change (or not) over the course of treatment, and document their associations with mental illness and comorbidity symptom severity, and to (2) obtain preliminary estimates on the predictability of treatment adherence and treatment outcome using features extracted from smartphone sensing and module engagement data over an early phase (e.g., the first two weeks) of the treatment regimen.
This pilot study will position us well to document feasibility, explore the utility of using mobile sensing to detect treatment-related behavioral changes, and provide preliminary evidence that could support a future large-scale clinical trial.
Status of this project: actively seeking students and collaborators
Preliminary data: multiple smartphone sensing and health assessment datasets from participant cohorts of 100-1000 individuals lasting 3-10 weeks.