Developing a methodology for using machine learning techniques to follow online mental health and well-being discourse among young people and adolescents to inform programming

 
 

The Issue

Approximately half of all poor mental health conditions manifest by early adolescence and available evidence suggests that around 10 to 20 percent of children and adolescents experience one of these conditions. Lack of timely interventions to address young people and adolescent mental health conditions can lead to limiting opportunities to lead fulfilling lives as adults and can be a contributing factor to premature mortality. It is key that organisations that deliver support services to adolescents understand what help is needed most and how to best deliver it.

Events like the COVID-19 pandemic, displacement and conflict prove difficult for those organisations to conduct relevant studies and gather information without digital means. The extent of young people’s engagement with the digital world creates an opportunity to deepen the understanding of the most up-to-date discourse amongst them on issues of mental health. The Data for Children Collaborative has previously conducted research on ‘Prevention of adolescent mental health conditions: is technology a possible source for good?’ which has helped to map out the existing landscape in this area. However, serious data gaps remain around how they discuss and engage with topics of mental health via social media and internet searches. More needs to be done to understand how this information can provide insights that can inform adequate and effective support, services and care.

Why Does it Matter?

Article 24 of the UN Convention on the Rights of the Child states that every child has the right to the best standard of health, including the promotion of mental health and wellbeing of children. Under this article, it is important to provide early identification and intervention for children and young people with mental health issues and to ensure that children can access a range of services promoting their mental health and wellbeing.

Sustainable Development Goal 3 also sets out to reduce premature mortality from non-contagious conditions through means including the promotion of mental health and wellbeing. Physical access for data collection and the time frame for methodology development, data collection, processing, and analysis have become significant barriers. The challenges posed during the pandemic emphasized the need to expand traditional research methods and incorporate non-traditional data sources such as big data into seeking insights.

Social media data can inform decision-making and policy adjustments to address mental health issues among adolescents, a topic that still carries stigma and for which sufficient support and service data are lacking. The Internet, as a powerful tool for reaching adolescents, presents opportunities for cost-effective interventions and research. The success of such initiatives can be replicated in other countries and applied to various adolescent-related topics. Understanding the digital footprint of adolescents can inform strategies to improve mental health care and reduce stigma.

Public and third sector analytics can benefit from big data, especially in predicting beneficiary behaviour and needs when data collection is challenging, and circumstances are constantly changing. UNICEF's work on mental health in the Europe and Central Asia region includes support for government institutions and civil society organizations providing online mental health services to adolescents.

In this challenging context, the development of a methodological tool becomes essential to complement evaluations regarding relevance, particularly as circumstances change. Analysing whether interventions align with current needs and problems is crucial for informed decision-making and policy adjustments. Ultimately, ensuring that interventions address present needs and problems is of paramount importance, regardless of their effectiveness, efficiency, or coherence. This information assists policymakers in determining the appropriateness of interventions in a rapidly changing landscape.

Our Project

Our nine-month project builds upon pre-existing work carried out by UNICEF in Ukraine in 2021-22 to use machine learning techniques to follow online mental health and well-being discourse among young people and adolescents. Our aim in this project is to take a more participatory approach with adolescents and young people. We want to work with young people to understand their use of social media and internet searches on mental health, the cultural language they use and the online contexts in which they seek mental health support and information. Engaging young people from Tajikistan and Kazakhstan will aid a better understanding of online discourse around mental health among adolescents in those regions.

The project will develop a transferable process model, including participatory data gathering and technical tool selection and adjustments, allowing the building of an analytical tool that will provide timely insights into adolescent mental health at a national scale. The overall methodology of how to build such a decision-support tool will be developed in a way that it can be adopted by various UNICEF country and regional offices with local adaptation, ultimately aiming to support UNICEF in creating tools to inform policy and prioritise relevant mental health interventions.

This project will:

1.       Use machine learning techniques to develop an analytical tool that enables a timely population-level view of current adolescent discourse on mental health.

2.       Work with UNICEF partners to build capacity and understanding of this analytical tool and the methods used through training materials such that the digital mental health data outputs can inform programming and policy developments.

3.       Produce a full methodological report on how to engage young people in different contexts to inform the creation of annotated datasets for use by this analytical tool. This report will underpin the ability to replicate research methods and enable adoption and adaptation by other international UNICEF offices to drive appropriate interventions and policy change.

 

Theme

Mental Health

Who is involved

UNICEF Europe & Central Asia Regional Office

UNICEF Kazakhstan Country Office

UNICEF Tajikistan Country Office

Data for Children Collaborative

The University of Edinburgh - School of Social & Political Science, School of Informatics and the School of Health & Social Science

 

Our Outputs

Previous
Previous

Understanding and overcoming barriers in data and information sharing relating to care experienced children, across public sector agencies and organisations in Scotland.

Next
Next

Building Heights Phase 2 - Calculating Line-Of-Sight Between a School Location and Radio Antennae Using Deep Learning Models on Satellite Data