Conversational agents and virtual reality are two emerging technologies that are increasingly being explored in mental health research. Although the combination of these technologies could provide easily accessible and cost-effective treatment for a wide range of health behaviors, use of this opportunity in healthcare has not yet been undertaken. Inherent difficulties of leveraging conversational agent solutions for non machine learning(ML) experts, and challenges - operational as well as legislative -regarding the exploitation of its capabilities, has posed a bottleneck for further adoption in the domain. This study applies interactive machine teaching principles, in the creation of a technical system enabling health-care clinicians (or any other non-ML experts) continuously develop and improve conversational agents deployed in VR applications. The methodology adopted first conducted a literature review providing background knowledge of the field, how conversational agents in VR could support clinicians and patients, methodological aspects and focus of the study. Secondly, a proof-of-concept VR system implementing a conversational agent for stress rehabilitation was developed in joint consultation with psychologists and later evaluated in an experimental study. This study presents a novel approach for deploying VR-based health interventions implementing conversational agents, and findings suggest the feasibility of such applications and the opportunities it has to support clinicians and patients in reaching their goals.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:umu-191709 |
| Date | January 2022 |
| Creators | Häggström Fordell, Vidar |
| Publisher | Umeå universitet, Institutionen för datavetenskap |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | UMNAD ; 1303 |
Page generated in 0.013 seconds