A Mixed-Methods Study of the Ethical Issues Surrounding Mobile Sensing in Digital Mental Health Interventions

University Of Virginia, Charlottesville VA

Investigators

Linked publications, trials & patents

Abstract

Abstract There is a crucial need for scalable, accessible mental health treatments that are delivered outside of clinics and can be integrated into daily life. Just-in-Time Adaptive Interventions (JITAIs) delivered via smart- phones and wearables represent a promising method to increase access to cost-effective and acceptable mental health care, and tailor in-the-moment interventions to best match the specific context of the individ- ual and their personal stressors, and determine when the individual is most likely to benefit from the inter- vention. However, these more scalable, personalized, context-sensitive interventions raise ethical questions tied to patient consent and acceptability, intrusiveness of monitoring, privacy and data security tied to both the collection of sensitive data and the associated computational methods applied. The main objective of this supplement is to answer these crucial ethical questions related to leveraging mobile sensing, wear- ables, and computational methods for socially anxious individuals in the context of digital mental health interventions. To our knowledge, this work is the first to tackle the specific ethical concerns for conducting research with mobile sensing devices with individuals high in social anxiety, which is especially important given this population has strong fears of evaluation and self-consciousness. To address this need, in Study 1, N=20 individuals high in trait social anxiety will be invited to engage in 1-hour, one-on-one, in-person interviews. We will follow a semi-structured interview guide to enquire about views toward wearable sen- sors (particularly watches) and passive monitoring on mobile phones. In Study 2, we will do a secondary data analysis, leveraging data collected from a group of socially anxious (N=46) participants as part of our R01 parent grant. Specifically, we will use existing data to empirically investigate the risk of person re-identification (i.e., predicting a participant's identity using features extracted from sensed data streams). More broadly, we will identify the key factors (e.g., individual, modalities/sensors, and contextual differ- ences) that contribute to privacy risks in biobehavioral data. After identifying the key factors contributing to privacy risks, we will develop and test privacy-preserving techniques that are tailored to those factors, to minimize the privacy risks while still obtaining valuable insights from the data. For JITAI's to meet their promise, we need to understand socially anxious individuals' preferences about different types of mobile sensing, various combinations of data streams, risks of sharing the resulting data with different people, and trade-offs between privacy on the one hand and knowledge production and clinical efficacy on the other hand. Together, this work will inform guidelines for researchers and practitioners so they can benefit from mobile sensing and JITAIs in a way that is ethical and patient-centered, while not compromising model performance.