Table 4.
Evaluation outcomes of mental health CAsa for youth (N=39).
| Dimensions and codes | Value, n (%) | Key trends | |||
| Dependent measures | Most studies focused on evaluating the effectiveness and acceptability of CAs | ||||
|
|
Effectiveness | 17 (44) |
|
||
|
|
Acceptability | 17 (44) |
|
||
|
|
Usability | 14 (36) |
|
||
|
|
User engagement | 10 (26) |
|
||
|
|
Validity or accuracy | 2 (5) |
|
||
|
|
Personalization | 1 (3) |
|
||
| Strengths | Easy access to useful mental health information communicated through friendly, empathetic, and humanlike responses was the major strength of mental health CAs | ||||
|
|
Accessibility | 14 (36) |
|
||
|
|
Useful content | 14 (36) |
|
||
|
|
Interactivity or engaging | 12 (31) |
|
||
|
|
Anonymity or confidentiality | 7 (18) |
|
||
|
|
Empathetic or friendly responses | 6 (15) |
|
||
|
|
Easy to navigate | 4 (10) |
|
||
|
|
Humanlike interaction | 2 (5) |
|
||
|
|
Nonjudgmental | 2 (5) |
|
||
| Weaknesses | Limited or repetitive content, lack of human language understanding, and robotic responses were the critical weaknesses of mental health CAs | ||||
|
|
Limited responses or content | 13 (33) |
|
||
|
|
Lack of natural language understanding | 8 (21) |
|
||
|
|
Lack of personalized content | 7 (18) |
|
||
|
|
Non-humanlike traits | 7 (18) |
|
||
|
|
Textual information with jargon | 5 (13) |
|
||
|
|
Inaccurate responses | 5 (13) |
|
||
|
|
Data privacy and confidentiality | 5 (13) |
|
||
| Ethical considerations | Ethical considerations such as privacy, confidentiality, and safety of CAs were addressed in a few studies | ||||
|
|
Privacy and confidentiality | 4 (10) |
|
||
|
|
Safety | 1 (3) |
|
||
aCA: conversational agent.