Expert Consensus Best Practices for the Safe, Ethical, and Effective Design and Implementation of Artificially Intelligent Conversational Agent (i.e., Chatbot/Virtual Human) Systems in Health Care App
Authors: Rizzo A, Mozgai S, Sigaras A, Rubin JE, Jotwani R
Journal: Journal of medical extended reality
mental health
psychology
open access
Abstract
Substance use disorders (SUDs) among veterans constitute a costly and complex public health challenge, compounded by high rates of comorbidities like post-traumatic stress disorder (PTSD) and traumatic brain injury (TBI). Psychotherapy such as the cognitive behavioral therapy (CBT)-based Transcending Self Therapy (TST) has shown promise in improving SUD treatment outcomes at VA Medical Centers. TST is a form of CBT. In addition to helping patients change maladaptive thoughts and behaviors, TST contends that it is important for patients to identify their moral compass and how to live life in accordance with their morals. Patients being treated with TST are encouraged to understand the thinking and behavior patterns that interfere with a value-oriented life as well as what they need to do to develop healthy interpersonal relationships and a passionate pursuit/meaningful life. The success of TST in its group and individual talk therapy formats led to an interest in investigating other avenues for delivering TST to patients. To this end, our previous work examined the feasibility of expanding TST into an immersive virtual reality (VR) environment Transcending Self Therapy Virtual Reality (TST-VR) to provide self-guided therapeutic engagement and immersive psychoeducation for veterans undergoing residential SUD treatment (). We did so by following a clinical VR development framework (the VR-CORE Model) that has three phases: VR1, VR2, and VR3. In this framework, VR1 studies focus on content development through human-centered design with patients and providers, VR2 trials conduct feasibility/acceptability/tolerability testing and initial clinical efficacy, and VR3 trials are randomized controlled studies or demonstration projects evaluating efficacy. Applying this framework, our initial feasibility study of a single Module prototype (VR1) obtained feedback that informed initial design and iterative refinements to TST-VR, wherein pilot participants reported positive reactions to the virtual environment. Based on this promising initial feedback, the program was expanded from one to four Modules of psychoeducation and activity content, and a VR2 study was undertaken. During this VR2 study, the program was further expanded from 4 Modules to 10 Modules. The initial 4-Module version focused mainly on core CBT concepts such as catching, checking and changing thoughts. For the 10-Module version, several existing videos were rewritten and reshot based on feedback, and new Modules were added around the following topics: New Beginnings, The Moral Compass, Relapse and Chain Analysis, Anxiety, Anger, and Behavioral Activation. The 10-Module version also incorporated ongoing user and clinician feedback to allow for smoother interactive gameplay. Changes were also added to support all levels of users with refined instructions and improved and more detailed feedback if the program detected that a user misunderstood instructions.