120 - Harnessing real world behavior data to optimize treatment delivery

Chair(s)

• 

  Jonah Meyerhoff, Ph.D.

  Research Assistant Professor
  Northwestern University
  Chicago, Illinois
• CS

  Caitlin Stamatis, Ph.D.

  Clinical Intern
  Northwestern University Feinberg School of Medicine
  Jersey City, New Jersey

Discussant(s)

• BT

  Bethany Teachman, Ph.D.

  Professor
  University of Virginia
  Charlottesville, Virginia

Presenter(s)

• AP

  Abhishek Pratap, Ph.D.

  CAMH
  Toronto, Ontario, Canada
• AS

  Andrew Sherrill, Ph.D.

  Emory University
  Atlanta, Georgia
• SM

  Susan Murphy, Ph.D.

  Harvard University
  Cambridge, Massachusetts
• CS

  Caitlin Stamatis, Ph.D.

  Clinical Intern
  Northwestern University Feinberg School of Medicine
  Jersey City, New Jersey

Mental health treatment gaps are growing worldwide, and rates of in-person mental health service use have not kept pace with rising rates of mental health conditions, particularly during the pandemic. One key avenue to increase access to care is through digital mental health interventions. These interventions can be delivered via online-connected devices, are effective for reducing symptoms in myriad disorders, and overcome many of the structural and attitudinal barriers that delay treatment for common mental health conditions. Moreover, these tools are low cost, can be used in private, and are delivered on-demand. However, the biggest challenge facing the success of digital mental health interventions remains low rates of sustained engagement, with the vast majority of digital mental health tools abandoned within the first 2 weeks of initiating use. Promoting engagement requires bridging gaps in fit between a digital mental health intervention’s offerings and an individual’s specific needs. 

Sensors embedded within digital devices offer promise to help customize, tailor, and deliver digital mental health interventions that are matched to a users’ specific needs, treatment targets, and availability, offering a potential solution to the engagement problem. Passively-sensed behavioral changes are associated with symptoms of mental health conditions; for instance, a reduction in activities and movement through space has been linked with anhedonia in people with depression, and psychophysiological changes may signal increased psychomotor agitation in individuals with generalized anxiety disorder. Promising but less studied areas include how the sentiment in text messages may signal changes in affective states, and how online search data may indicate risk. Better understanding links between such passively-sensed data and psychological symptoms offers opportunities to increase the fit between a digital mental health intervention and a user’s needs via just-in-time, idiographic interventions. 

In this panel, we will hear from a series of speakers on how passively sensed data can be used to optimize treatment delivery, improve symptom detection, and enhance engagement. The first two talks will highlight the impact of personal sensing and machine learning methods in clinical trials. Presenter 1 will share results from a trial testing the use of sensor data during prolonged exposure in an applied clinical treatment setting for PTSD. Presenter 2 will discuss statistical methods for determining the impact of artificial intelligence-based treatment personalization on clinical outcomes. The second two talks will focus on text and language-based personal sensing to improve risk prediction. Presenter 3 will share novel findings on the assessment of self-harm risk based on online search data. Presenter 4 will discuss the differential prediction of depression and anxiety symptoms based on in-vivo text messaging data. The Discussant, a leading expert in digital mental health, will synthesize findings to highlight how passively sensed data can enhance personalized treatment, improve access to care, and promote the scalability of evidence-based care in the context of mental health treatment gaps that have widened during the pandemic.

Presentations:

• (Sym 120) Understanding Real-world Personalized Risk Factors of Self-harm Through Online Information-seeking Behavior of Individuals
  

  Presenter: Abhishek Pratap, Ph.D. – CAMH

  Co-author: Patricia Areán, Ph.D. – Department of Psychiatry and Behavioral Sciences, University of Washington

  Co-author: Honor Hsin, MD PHD – Kaiser Permanente

  Co-author: Patrick J Heagerty, PhD – Univ of Washington

  Co-author: Trevor Cohen, MBChB, PhD, FACMI – University of Washington

  Co-author: Courtney Bagge, PhD – University of Michigan

  Co-author: Katherine Comtois, PhD, MPH – University of Washington
• (Sym 120) Ubiquitous Computing Can Enhance Patient Engagement, Patient-clinician Communication, and Clinician Decision-making: Preliminary Findings of the Prolonged Exposure Collective Sensing System
  

  Presenter: Andrew M. Sherrill, Ph.D. – Emory University

  Co-author: Barbara Rothbaum, Ph.D., ABPP – Emory University School of Medicine

  Co-author: Sheila Rauch, ABPP, Ph.D. – Emory University School of Medicine

  Co-author: Hayley Evans, PhD – Georgia Institute of Technology

  Co-author: Rosa Arriaga, PhD – Georgia Institute of Technology
• (Sym 120) Assessing Personalization in Digital Health
  

  Presenter: Susan Murphy, Ph.D. – Harvard University
• (Sym 120) Prospective Associations of Text-message-based Sentiment with Symptoms of Depression, Generalized Anxiety, and Social Anxiety
  

  Presenter: Caitlin A. Stamatis, Ph.D. – Northwestern University Feinberg School of Medicine

  Co-author: Jonah Meyerhoff, Ph.D. – Northwestern University

  Co-author: Tingting Liu, Ph.D. – Technology & Translational Research Unit, National Institute on Drug Abuse (NIDA IRP), National Institutes of Health (NIH)

  Co-author: Garrick Sherman, Ph.D. – Department of Computer Science, University of Pennsylvania

  Co-author: Harry Wang, B.S. – Department of Computer Science, University of Pennsylvania

  Co-author: Tony Liu, M.S. – Department of Computer Science, University of Pennsylvania

  Co-author: Brenda Curtis, Ph.D., MsPH – Technology & Translational Research Unit, National Institute on Drug Abuse (NIDA IRP), National Institutes of Health (NIH)

  Co-author: Lyle Ungar, Ph.D. – Department of Computer Science, University of Pennsylvania

  Co-author: David Mohr, PhD – Northwestern University