Symposia
Technology
Andrew M. Sherrill, Ph.D.
Emory University
Atlanta, Georgia
Barbara Rothbaum, Ph.D., ABPP
Professor in Psychiatry
Emory University School of Medicine
Atlanta, GA
Sheila Rauch, ABPP, Ph.D.
Professor in Psychiatry
Emory University School of Medicine
Atlanta, GA
Hayley Evans, PhD
Research Affiliate
Georgia Institute of Technology
Atlanta, Georgia
Rosa Arriaga, PhD
Associate Professor
Georgia Institute of Technology
Atlanta, Georgia
Implementation of cognitive behavioral therapies (CBTs) is often constrained by the use of clinical data that are subjective, retrospective, unreliable, and narrow. Clinicians and patients are often unclear if the patient is sufficiently engaging in therapeutic exercises and if the assigned exercises are resulting in desired changes. Clinicians would benefit from the collection of several continuous, noninvasive, and objective data streams to efficiently track and respond to their patients’ needs. Similarly, patients would benefit from timely and objective feedback about how to improve their engagement in therapeutic exercises. To address these needs, we (1) developed a user-tailored sensing system that allows patient data transfer and information extraction during therapeutic exercises during and between sessions and (2) designed system interfaces to facilitate continuous monitoring for both clinicians and patients and to enable efficient clinician-patient communication at the point of care. Our first use case of this computational toolkit is prolonged exposure (PE) therapy for PTSD. This system, called the PE Collective Sensing System (PECSS), includes a mobile app for patients and an online dashboard for clinicians. PECSS includes active data collection (e.g., ratings of distress during exposures) and four streams of passive data collection: mobile sensors (e.g., location, movement, acoustic environmental signals, use of other mobile apps), physiological sensors (e.g., heart rate), social sensors (e.g., targeted SMS with trusted others), and language sensors (e.g., natural language processor during imaginal exposure). Our presentation will first describe the iterative process of designing PECSS using methods from human-computer interaction. Our presentation will then include pilot data from the first deployment of PECSS, namely with military veterans diagnosed with PTSD receiving care at Emory Healthcare Veterans Program. We will discuss next research steps including the develop of computational models using PECSS inputs to provide individualized patient feedback and automated suggestions to facilitate clinical decision-making.