Personalized AI-Driven Models in Cognitive Behavioral Therapy for Anxiety

Untreated anxiety undermines long-term physical and emotional wellbeing, especially among college students, with rates worsening since the onset of the COVID-19 pandemic. Cognitive Behavioral Therapy (CBT) is the leading evidence-based intervention for anxiety, but many students fail to complete exercises between CBT sessions, reducing its effectiveness. Socially assistive robots (SARs) help promote adherence to home-based practice in the context of elder care, social skill learning, and physical therapy, but it is unknown how SARs can enhance CBT. The specific objective of this research is to develop personalized CBT SARs that can support CBT compliance for college students with anxiety. To meet the goals of the proposed work, these studies will determine how SAR personalization based on implicit and explicit feedback can help promote greater CBT compliance and anxiety reduction outcomes for students.

The investigators will conduct a Phase 1 parallel intervention study with random assignment. The investigators will randomly assign consented college students reporting anxiety symptoms to one of two 6-week Cognitive Behavioral Therapy (CBT) exercise conditions that were validated in our preliminary studies. The participant will be randomly assigned to a condition (personalized versus not personalized socially assistive robot (SAR) CBT) and will be informed of their condition by the investigators when the participant is ready to start their home practice with the SAR CBT companion. During the initial baseline in-home visit, participants will be provided with a SAR and instructions for how to complete their SAR CBT exercises daily for 6 weeks. During the 6-week home-based study period that follows the baseline visit, the participant will be reminded daily to complete a CBT exercise with their SAR. The investigators will be the point of contact via phone or email for all participants who need assistance (e.g., with SAR/software issues). The investigators will evaluate personalized re-engagement strategies derived from implicit feedback from individuals with clinically elevated anxiety. This will be done through an in-home between-subjects user study involving 60 student participants who will be randomly assigned to either a control condition without re-engagement feedback or a personalized feedback condition with personalized re-engagement feedback. Quantitative behavioral and subjective metrics will be compared between the conditions to assess the impact of personalized feedback on feelings of relatedness with the robot, intrinsic motivation, engagement, and adherence, guided by self-determination theory. The study aims to determine whether personalized re-engagement feedback enhances the intrinsic motivation, engagement, and adherence when using the SAR companion to deliver CBT concepts to individuals with clinically elevated anxiety. The second aim is to evaluate the personalization of robot attributes based on explicit user feedback in individuals with clinically elevated anxiety. This will be done through an in-home between-subjects user study where 60 participants are randomly assigned to either a control condition without the capability to personalize the robot's attributes or a personalization condition where the participant can modify the robot's attributes at the start of the study and between CBT sessions. This study aims to determine whether granting users the ability to choose the robot's attributes will lead to higher levels of intrinsic motivation, engagement, and adherence during interaction with the SAR companion.