AI-powered Human Digital Twins in Virtual Therapeutic Sessions

Abstract

When patients are unwilling to speak openly, therapists miss valuable information and may not be able to provide best possible care. This study investigated the idea of human Digital Twins (DT) of patients and therapists as a novel solution to encourage them to speak openly. We proposed a methodology to create highly realistic human DT using computer vision, 3D scanning, Metahumans tool, and Unreal game engine. Based on this methodology, we created two versions of virtual therapeutic session: a) video-based virtual sessions with real human and b) virtual sessions with DT of patients and therapists. Twelve therapists then were asked to evaluate technology acceptance of each virtual session. Participants reported significantly lower trust to virtual sessions with highly realistic digital human compared to video-based one. Findings from this study may provide insights regarding opportunities and challenges of virtual therapy using highly realistic digital human technologies. This work also contributes to the virtual therapeutic research field by proposing an AI-powered methodology to create human DT of patients and therapists. Further studies are needed to provide more validation of the efficiency of this approach in virtual sessions for mental health support.