[Basic, Clinical and Social Perspectives on the Use of Virtual Characters in Mental Health].

Along other breakthroughs in computer sciences, such as artificial intelligence, virtual characters (i.e. digitally represented characters featuring a human appearance or not) are foreseen as potential providers of mental healthcare services. However, their current use in clinical practice is marginal and limited to an assistive role to help clinicians in their practices. Safety and efficiency concerns, as well as a general lack of knowledge and experience, may explain this discrepancy between the expected (sometimes futuristic) and current use of virtual characters. An overview of recent evidence would help pinpoint the main concerns and challenges pertaining to their use in mental healthcare. Objective This paper aims to inform relevant actors, including clinicians, on the potential of virtual characters in mental healthcare practices and to raise awareness on societal challenges regarding their use. Method A narrative literature review was conducted to summarize basic and clinical research findings, and to outline an in-depth discussion on various societal caveats related to the inclusion of virtual characters. Results Basic studies highlight several characteristics of the virtual characters that seem to influence patient-clinician interactions. These characteristics can be classified into two categories: perceptual (e.g. realism) and social features (i.e. attribution of social categories such as gender). To this day, many interventions and/or assessments using virtual characters have shown various levels of efficiency in mental health, and certain elements of a therapeutic relationship (e.g. alliance and empathy) may even be triggered during an interaction with a virtual character. To develop and increase the use of virtual characters, numerous socioeconomic and ethical issues must be examined. Although the accessibility and the availability of virtual characters are an undeniable advantage for their use in mental healthcare, some inequities about their application remain. In addition, the accumulation of biometric data (e.g. heart rate) could provide valuable information to clinicians and could help develop autonomous virtual characters, which raises concerns over issues of security and privacy. This paper proposes some recommendations to avoid such undesirable outcomes. Conclusion Due to their promising features, the inclusion of virtual characters will no doubt be increasingly prevalent in mental healthcare services. All involved actors should thus be informed about specific challenges raised by such breakthroughs. They should also actively participate in discussions regarding the development of virtual characters in order to adopt unified recommendations for their safe and ethical use in mental healthcare.