Artificial intelligence algorithms promoting help-seeking behavior among depressed youth

Kim Kristoffer Dysthe

Keywords: Chatbot algorithm. Help-seeking behavior. Youth depression.


Symptoms of depression are common in adolescents. If not treated, symptoms can develop into more severe mood disorders later in life, leaving a window of opportunity for intervention in general practice. However, depressed adolescents often avoid seeking professional help. Instead, they search for information on the Internet, often being misleading and lacking scientific evidence. From previous research, we know that adolescents are positive to new clinical designs using technology as an integrated part of therapy, but we know little about the information needs of youth with symptoms of depression. This study aims to establish the empirical background needed to develop mental health literacy interventions promoting positive help-seeking attitudes.

Research questions:

What are the information needs of youth with symptoms of depression, and what information will affect help-seeking behavior?


Our dataset consists of 15 000 questions about mental health problems written by youth aged 16-25 years on Internet services. We use text-mining tools for content analysis to find texts describing symptoms of depression. We conduct a qualitative study within a realist epistemology, using thematic analysis based on the Health beliefs model, asking what kind of information impedes, and what enables help-seeking behavior. From the data, we extract dialogues based on relevant questions and corresponding answers.



Being part of the larger multi-center research project “Social Health Bots,” we will use the dialogues to train chat-BOT algorithms to detect and categorize questions related to help-seeking behavior. Internet or mobile clinical applications based on such algorithms can advise depressed youth to seek professional assistance. Beyond the scope of this study, further interaction - and service design, as well as clinical trials, are necessary for validation and utilization of technology in general practice.

Points for discussion:

How we can use data from Internet forums in health research.

The implementation of technology into clinical practice - as simple as it seems? Design process and research possibilities.