Category: Technology
Mohr, D. C., Zhang, M., & Schueller, S. M. (2017). Personal Sensing: Understanding Mental Health Using Ubiquitous Sensors and Machine Learning. Annu Rev Clin Psychol, 13, 23-47. https://doi.org/10.1146/annurev-clinpsy-032816-044949
,Torous, J., Kiang, M. V., Lorme, J., & Onnela, J. P. (2016). New tools for new research in psychiatry: A scalable and customizable platform to empower data driven smartphone research. JMIR Ment Health, 3(2), e16. https://doi.org/10.2196/mental.5165
,Zarate, D., Stavropoulos, V., Ball, M., de Sena Collier, G., & Jacobson, N. C. (2022). Exploring the digital footprint of depression: a PRISMA systematic literature review of the empirical evidence. BMC Psychiatry, 22(1), 421. https://doi.org/10.1186/s12888-022-04013-y
,Orsolini, L., Fiorani, M., & Volpe, U. (2020). Digital Phenotyping in Bipolar Disorder: Which Integration with Clinical Endophenotypes and Biomarkers? Int J Mol Sci, 21(20). https://doi.org/10.3390/ijms21207684
Jessica Lipschitz, Ph.D. (she/her/hers)
Associate Director, Digital Behavioral Health & Informatics Research Program/Assistant Professor
Brigham and Women’s Hospital / Harvard Medical School
Boston, Massachusetts
Stephen Schueller, Ph.D. (he/him/his)
Associate Professor
University of California, Irvine
Irvine, California
Caitlin Stamatis, Ph.D. (she/her/hers)
Research Fellow
Northwestern University Feinberg School of Medicine
Chicago, Illinois
Jessica Lipschitz, Ph.D. (she/her/hers)
Associate Director, Digital Behavioral Health & Informatics Research Program/Assistant Professor
Brigham and Women’s Hospital / Harvard Medical School
Boston, Massachusetts
Kelly Zuromski, Ph.D. (she/her/hers)
Research Associate
Harvard University
Cambridge, Massachusetts
Digital tools which provide information about patient symptoms, behavior and functioning in real-time have sparked substantial interest over the past decade. Early on, such practices involved digital progress monitoring for routine appointments and ecological momentary assessment for higher-frequency observation of key symptoms. Now, such practices are evolving to incorporate digital phenotyping (Onnela & Rauch, 2016; Torous et al., 2016) or personal sensing (Mohr et al., 2017)—the moment-to-moment quantification of the individual lived experience, in situ, using data from personal digital devices, such as smartphones and wearables. These devices can unobtrusively collect an array of behavioral and biological information that has been found to be associated with mental health concerns, including sleep patterns, daily activity, heart rate, keyboard typing dynamics, voice data, geolocation, temperature and call/text message logs (Orsolini et al., 2020; Zarate et al., 2022).
These methods could transform the behavioral health treatment paradigm in several ways. First, they could facilitate objective and low-burden progress-monitoring, allowing clinicians to modify treatment plans as needed. Second, they could serve as a safety net for high-risk patients who may deteriorate rapidly between sessions, alerting clinicians to needed outreach or crisis management protocols. Finally, they could inform adjunctive digitalized treatments responsive to patients’ ongoing symptoms, such as just-in-time adaptive interventions (JITAIs).
Some digital tools have become commonplace (e.g., telehealth), but passive digital monitoring of real time behavioral signals is rarely integrated into care. Still, it is possible today and featured in some clinics. As these methods advance, they will likely become part of evidence-based behavioral health treatment in the near future. As such, clinicians need to be aware of the evidence and drawbacks of these approaches.
The three studies discussed in this symposium illustrate the promise of these methods and the current state of prediction accuracy across several important clinical populations. First, we will present findings from a large-scale study of smartphone-based passive sensing to identify distal and proximal markers of depression and anxiety. Second, we will discuss a study that used activity, sleep and heart rate data from Fitbit devices to detect occurrence of depression and mania in patients with bipolar disorder. Last, we will review a study examining the relationship between both objective, passively collected and subjectively reported sleep data and suicidal intent in a large sample of adolescents and adults during the high-risk, post-hospitalization window.
All three studies use consumer personal digital devices and show predictive capacity for key clinical features of the population of interest. They also all illustrate where these technologies and predictive algorithms still require additional validation. Discussion will focus on best practices when using these methods in research, how these methods show promise to complement standard care, and which aspects of these methods are and are not ready for real-world implementation.
