Infographic: Developing a methodology for using AI to identify social media discussions on mental health and well-being

As part of our project focusing on developing a methodology for using AI to identify social media discussions of mental health and well-being, we ran a few youth engagement sessions together with UNICEF country offices in Kazakhstan and Tajikistan.

What did we do?

• Involved Youth

  We asked young people in Kazakhstan and Tajikistan how and where they talk about mental health online.

  We wanted to understand the specific language they use and the topics they’d explore.

• Gained Language Insights

  Based on what young people shared, we made a list of mental health keywords and commonly used language specific to young people.

  We used these lists to instruct a type of artificial intelligence (AI) - a large language model, to find social media posts by young people about mental health.

• Validated Results

  We checked the results from the AI by asking young people to look at real social media posts and tell us if they were about mental health and written by other young people.

How can young people’s input continue to have an impact?

• Continued Participation

  Youth Participation is key to understanding the language used by young people in general and on a specific topic. Humans need to validate results delivered by AI.

• Better Support

  The methods developed in this project could guide strategies for improving mental health services and reducing stigma. They can also support timely interventions.

• Wider Application

  The methods developed can also be applied to many different social media sites in various countries and in other languages.

---