The development of a World Health Organization transdiagnostic chatbot intervention for distressed adolescents and...

Common mental disorders are prevalent in young people in low- and middle-income countries (LMICs). Digitally delivered interventions have the potential to overcome many structural and psychosocial barriers to mental health care. Chatbots have been proposed as one potentially acceptable and feasible method that may increase engagement. Yet, there is currently limited evidence for their efficacy in reducing psychological distress. This paper summarises the development of a World Health Organization digital psychological intervention for young people experiencing impairing psychological distress, developed in line with Human Centred Design (HCD) principles.

This study refined and adapted a chatbot intervention initially developed for adolescents aged 15-18 years that was completed in consultation with end-users in this age group (N = 236), community members (N = 73), and psychology intervention experts (N = 9) across varied settings. The purpose was to create an adaptation fit for use by young adults aged 18-21 years experiencing psychological distress in Jordan.

The current study followed a limited user-centred design process involving focus groups and key informant interviews with stakeholders including young adults aged 18-21 years (N = 33), community members (N = 13), and psychology intervention experts (N = 11). Iterative design development occurred throughout the cultural adaptation and refinement process.

There was a clear preference for a chatbot based intervention that included interactions with fictional characters with relatable problems. The chatbot content followed a transdiagnostic model that addressed common problems including low mood, stress and anger with reference to vocational, familial and interpersonal stressors that the target population commonly faced. It followed a non-AI decision tree format with multiple sessions and was designed to be adaptable for use in different countries with different populations and software systems. Prototype versions of the chatbot were well-received by adolescents (15-18-year-old) and young adults (18-21-year-old).

This is the first report of the development of a chatbot intervention for adolescents and young adults in LMICs that was designed using a HCD framework. Systematic end-user engagement through all phases of the research aimed to make this intervention acceptable and useable for adolescents and young adults in a wide variety of settings. The chatbot is currently being tested in randomised controlled trials in Jordan and Lithuania.