This chapter addresses the first aim to "map and analyze the current uses of conversational agents in healthcare and identify knowledge gaps by performing a scoping review." Section 4.1 provides some background information on existing reviews on healthcare CAs, what is missing from the literature, and the specific objectives of this review. The methodology is described in section 4.2 and the results are presented in section 4.3 both narratively and diagrammatically.The implications of the findings, strengths, and limitations of this review as well as suggestions for future research are outlined in section 4.4. This scoping review was conducted using the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) Checklist [123] and the updated Arksey and O'Malley framework [124]. To address the research questions presented in 4.1.1, a broad literature search needed to be conducted to identify the various usage applications of CAs. This required a comprehensive search of several databases using a multitude of synonyms. A broad literature search was conducted in April 2019 in MEDLINE (Medical Literature Analysis and Retrieval System Online; Ovid), EMBASE (Excerpta Medica database; Ovid), PubMed, Scopus, and Cochrane Central. Given the novelty of the field, and to increase comprehensiveness, gray literature was also searched for in the OCLC WorldCat database, ResearchGate, Google Scholar, OpenGrey, and the first 10 pages of Google. There was an extensive list of 63 search terms, which were various iterations of the term "conversational agents" (Appendix 4.1). These synonyms were generated using a web-based search and by identifying specific terms or phrases used in the titles of articles discussing health care conversational agents. Snowball searching of relevant articles and systematic reviews was also done to identity more articles related to the review. The PRISMA checklist for scoping reviews was completed and is presented in Appendix 4.2.
Inclusion and Exclusion CriteriaPrimary research studies that had evaluated a CA implemented for healthcare were included.Articles that just presented a proposal for CA development, articles that mentioned conversational agents briefly or as an insignificant part of a review, as well as opinion pieces and articles where primary research was not conducted, were excluded. A further point of exclusion was articles with poorly reported data on CA assessments where there was minimal or no evaluation data. In addition, articles concerning ECAs, relational agents, animated conversational agents, or other conversational agents with a visual or animated component were excluded. ECAs are computer-generated virtual avatars that interact with users with both verbal and non-verbal communication [125]. Relational agents are a type of ECA designed to create long-term deep and meaningful relationships with individuals [126]. As the user's interaction may be affected by nonverbal behaviors, graphics, and layout of the program, it was decided th...