Issues in Identifying Strategies for Youth Mental Well-Being in Stockholm Municipalities Using Participatory Sessions and Text Mining: Qualitative Study.

Background: Socioeconomic and environmental factors influence youth mental well-being. Promoting mental well-being is essential to support youths' development toward adulthood with good mental health. Different Stockholm municipalities have adopted strategies to promote youth well-being. However, contextualizing and perceiving goals and mechanisms at the local municipal level is difficult. Thus, comparing or tracking their conception, purpose, and characteristics has been challenging. Objective: We aimed to use data visualizations developed from a fusion of data sources to facilitate stakeholder conversations on promoting youth mental well-being within a municipality. We strive to demonstrate our methodology of using data visualizations as "boundary objects," which are cognitive artifacts that bridge knowledge from various domains to elicit understanding from specialized and siloed parts of a health delivery system. Methods: Stakeholders from the municipalities of Lidingö and Nynäshamn participated in the study. A total of 15 workshops were conducted: 6 with only Lidingö participants, 6 with only Nynäshamn participants, and 3 with mixed participants. The sessions were conducted via Microsoft Teams or as physical sessions in Swedish and lasted between 60 and 90 minutes. Interactions were recorded with consent from participants. Recordings were transcribed using Amberscript software. We used matrix factorization with Kullback-Leibler divergence to extract 1000 features and created 10 topic clusters with 20 top words. We used the identified words and phrases to backtrack within the transcripts and to identify dialogues where they were used. We summarized participants' interactions across all the workshops to identify factors or strategies discussed for youth well-being. Results: Participants noted that these sessions allowed them to contextualize their local observations from municipalities relative to the status of other municipalities in the national statistics. They indicated that they conceptualized well-being differently in their respective municipalities and between different professional backgrounds, and the sources of stress for youth differed. They noted the differences in the strategy and data collected for tracking youth well-being. Promotion of sports was a common strategy, while options for leisure activities differed between municipalities and professions. Conclusions: Based on our observations and analysis of the transcripts from participatory workshops, we observed that the data-driven visualizations helped stakeholders from different departments of Lidingö and Nynäshamn municipalities to identify and bridge knowledge gaps caused by data silos. Participants noted proposals to modify future surveys and identified that this approach to visualizations would help them to share knowledge and maintain a long-term and sustainable collaboration across departments.