Impact of a novel mental health chatbot on depression knowledge among youth with HIV: A pilot study

Introduction: Adolescents living with HIV (ALWH) face disproportionately high rates of depression and anxiety, which can negatively affect antiretroviral adherence, viral suppression, and overall health outcomes. Despite the close relationship between mental wellness and HIV care, mental health services remain inadequately integrated into HIV programs, particularly in low- and middle-income countries. To address this gap, we developed EVA (Educaci&oacuten, Vinculaci&oacuten y Autoayuda), a mental health chatbot co-designed with ALWH to provide psychoeducation on depression and anxiety, teach self-help skills, and facilitate linkage to mental health resources. We evaluated EVA's preliminary impact on depression knowledge and its feasibility and acceptability among ALWH in Lima, Peru. Methods: During January - August 2024, ALWH in Lima, Peru completed baseline assessments, engaged independently with EVA for 20 minutes, and completed endline measures. The primary outcome was change in depression knowledge measured using the Adolescent Depression Knowledge Questionnaire (ADKQ). Secondary outcomes included acceptability, appropriateness, feasibility, intention to use and recommend the chatbot, and satisfaction with the chatbot's features and content. Paired-samples t- tests were used to evaluate pre-post changes in depression knowledge. Results: Fifty ALWH aged 11 - 19 years participated, among whom 70% reported greater-than-minimal depressive symptoms, 60% greater-than-minimal anxiety symptoms, and 92% moderate or high perceived stress. Following a single interaction with EVA, depression knowledge increased significantly, from a mean ADKQ score of 6.88 (SD=1.85) to 8.12 (SD=2.20), t(49) = -5.03, p<.001, representing a large effect size (d=1.74). The chatbot was rated highly for acceptability, appropriateness, and feasibility, and participants reported strong intentions to use and recommend it, as well as high satisfaction with its educational content, usability, and self-help resources. Conclusions: A brief interaction with a mental health chatbot developed with and for ALWH in Peru significantly increased depression knowledge and was rated highly acceptable and feasible. Mental health chatbots may offer a low-cost, scalable approach to support mental health education and linkage to care within adolescent HIV services. Future longitudinal studies should assess sustained impact on mental health outcomes and help-seeking behavior; however, the benefits of chatbots should be weighed against their potential mental health risks.